linux/fs/f2fs/node.c

3234 lines
76 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0
/*
* fs/f2fs/node.c
*
* Copyright (c) 2012 Samsung Electronics Co., Ltd.
* http://www.samsung.com/
*/
#include <linux/fs.h>
#include <linux/f2fs_fs.h>
#include <linux/mpage.h>
#include <linux/backing-dev.h>
#include <linux/blkdev.h>
#include <linux/pagevec.h>
#include <linux/swap.h>
#include "f2fs.h"
#include "node.h"
#include "segment.h"
#include "xattr.h"
#include "trace.h"
#include <trace/events/f2fs.h>
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
#define on_f2fs_build_free_nids(nmi) mutex_is_locked(&(nm_i)->build_lock)
static struct kmem_cache *nat_entry_slab;
static struct kmem_cache *free_nid_slab;
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
static struct kmem_cache *nat_entry_set_slab;
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
static struct kmem_cache *fsync_node_entry_slab;
f2fs: give message and set need_fsck given broken node id 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=d154ec99402c6f628887 C reproducer: https://syzkaller.appspot.com/x/repro.c?id=5414336294027264 syzkaller reproducer: https://syzkaller.appspot.com/x/repro.syz?id=5471683234234368 Raw console output: https://syzkaller.appspot.com/x/log.txt?id=5436660795834368 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+d154ec99402c6f628887@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 ------------[ cut here ]------------ kernel BUG at fs/f2fs/node.c:1185! invalid opcode: 0000 [#1] SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) Modules linked in: CPU: 1 PID: 4549 Comm: syzkaller704305 Not tainted 4.17.0-rc1+ #10 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:__get_node_page+0xb68/0x16e0 fs/f2fs/node.c:1185 RSP: 0018:ffff8801d960e820 EFLAGS: 00010293 RAX: ffff8801d88205c0 RBX: 0000000000000003 RCX: ffffffff82f6cc06 RDX: 0000000000000000 RSI: ffffffff82f6d5e8 RDI: 0000000000000004 RBP: ffff8801d960ec30 R08: ffff8801d88205c0 R09: ffffed003b5e46c2 R10: 0000000000000003 R11: 0000000000000003 R12: ffff8801a86e00c0 R13: 0000000000000001 R14: ffff8801a86e0530 R15: ffff8801d9745240 FS: 000000000072c880(0000) GS:ffff8801daf00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f3d403209b8 CR3: 00000001d8f3f000 CR4: 00000000001406e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: get_node_page fs/f2fs/node.c:1237 [inline] truncate_xattr_node+0x152/0x2e0 fs/f2fs/node.c:1014 remove_inode_page+0x200/0xaf0 fs/f2fs/node.c:1039 f2fs_evict_inode+0xe86/0x1710 fs/f2fs/inode.c:547 evict+0x4a6/0x960 fs/inode.c:557 iput_final fs/inode.c:1519 [inline] iput+0x62d/0xa80 fs/inode.c:1545 f2fs_fill_super+0x5f4e/0x7bf0 fs/f2fs/super.c:2849 mount_bdev+0x30c/0x3e0 fs/super.c:1164 f2fs_mount+0x34/0x40 fs/f2fs/super.c:3020 mount_fs+0xae/0x328 fs/super.c:1267 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:2518 [inline] do_mount+0x564/0x3070 fs/namespace.c:2848 ksys_mount+0x12d/0x140 fs/namespace.c:3064 __do_sys_mount fs/namespace.c:3078 [inline] __se_sys_mount fs/namespace.c:3075 [inline] __x64_sys_mount+0xbe/0x150 fs/namespace.c:3075 do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x443dea RSP: 002b:00007ffcc7882368 EFLAGS: 00000297 ORIG_RAX: 00000000000000a5 RAX: ffffffffffffffda RBX: 0000000020000c00 RCX: 0000000000443dea RDX: 0000000020000000 RSI: 0000000020000100 RDI: 00007ffcc7882370 RBP: 0000000000000003 R08: 0000000020016a00 R09: 000000000000000a R10: 0000000000000000 R11: 0000000000000297 R12: 0000000000000004 R13: 0000000000402ce0 R14: 0000000000000000 R15: 0000000000000000 RIP: __get_node_page+0xb68/0x16e0 fs/f2fs/node.c:1185 RSP: ffff8801d960e820 ---[ end trace 4edbeb71f002bb76 ]--- Reported-and-tested-by: syzbot+d154ec99402c6f628887@syzkaller.appspotmail.com Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-04-24 05:02:31 +00:00
/*
* Check whether the given nid is within node id range.
*/
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_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid)
f2fs: give message and set need_fsck given broken node id 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=d154ec99402c6f628887 C reproducer: https://syzkaller.appspot.com/x/repro.c?id=5414336294027264 syzkaller reproducer: https://syzkaller.appspot.com/x/repro.syz?id=5471683234234368 Raw console output: https://syzkaller.appspot.com/x/log.txt?id=5436660795834368 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+d154ec99402c6f628887@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 ------------[ cut here ]------------ kernel BUG at fs/f2fs/node.c:1185! invalid opcode: 0000 [#1] SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) Modules linked in: CPU: 1 PID: 4549 Comm: syzkaller704305 Not tainted 4.17.0-rc1+ #10 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:__get_node_page+0xb68/0x16e0 fs/f2fs/node.c:1185 RSP: 0018:ffff8801d960e820 EFLAGS: 00010293 RAX: ffff8801d88205c0 RBX: 0000000000000003 RCX: ffffffff82f6cc06 RDX: 0000000000000000 RSI: ffffffff82f6d5e8 RDI: 0000000000000004 RBP: ffff8801d960ec30 R08: ffff8801d88205c0 R09: ffffed003b5e46c2 R10: 0000000000000003 R11: 0000000000000003 R12: ffff8801a86e00c0 R13: 0000000000000001 R14: ffff8801a86e0530 R15: ffff8801d9745240 FS: 000000000072c880(0000) GS:ffff8801daf00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f3d403209b8 CR3: 00000001d8f3f000 CR4: 00000000001406e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: get_node_page fs/f2fs/node.c:1237 [inline] truncate_xattr_node+0x152/0x2e0 fs/f2fs/node.c:1014 remove_inode_page+0x200/0xaf0 fs/f2fs/node.c:1039 f2fs_evict_inode+0xe86/0x1710 fs/f2fs/inode.c:547 evict+0x4a6/0x960 fs/inode.c:557 iput_final fs/inode.c:1519 [inline] iput+0x62d/0xa80 fs/inode.c:1545 f2fs_fill_super+0x5f4e/0x7bf0 fs/f2fs/super.c:2849 mount_bdev+0x30c/0x3e0 fs/super.c:1164 f2fs_mount+0x34/0x40 fs/f2fs/super.c:3020 mount_fs+0xae/0x328 fs/super.c:1267 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:2518 [inline] do_mount+0x564/0x3070 fs/namespace.c:2848 ksys_mount+0x12d/0x140 fs/namespace.c:3064 __do_sys_mount fs/namespace.c:3078 [inline] __se_sys_mount fs/namespace.c:3075 [inline] __x64_sys_mount+0xbe/0x150 fs/namespace.c:3075 do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x443dea RSP: 002b:00007ffcc7882368 EFLAGS: 00000297 ORIG_RAX: 00000000000000a5 RAX: ffffffffffffffda RBX: 0000000020000c00 RCX: 0000000000443dea RDX: 0000000020000000 RSI: 0000000020000100 RDI: 00007ffcc7882370 RBP: 0000000000000003 R08: 0000000020016a00 R09: 000000000000000a R10: 0000000000000000 R11: 0000000000000297 R12: 0000000000000004 R13: 0000000000402ce0 R14: 0000000000000000 R15: 0000000000000000 RIP: __get_node_page+0xb68/0x16e0 fs/f2fs/node.c:1185 RSP: ffff8801d960e820 ---[ end trace 4edbeb71f002bb76 ]--- Reported-and-tested-by: syzbot+d154ec99402c6f628887@syzkaller.appspotmail.com Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-04-24 05:02:31 +00:00
{
if (unlikely(nid < F2FS_ROOT_INO(sbi) || nid >= NM_I(sbi)->max_nid)) {
set_sbi_flag(sbi, SBI_NEED_FSCK);
f2fs_warn(sbi, "%s: out-of-range nid=%x, run fsck to fix.",
__func__, nid);
return -EFSCORRUPTED;
f2fs: give message and set need_fsck given broken node id 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=d154ec99402c6f628887 C reproducer: https://syzkaller.appspot.com/x/repro.c?id=5414336294027264 syzkaller reproducer: https://syzkaller.appspot.com/x/repro.syz?id=5471683234234368 Raw console output: https://syzkaller.appspot.com/x/log.txt?id=5436660795834368 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+d154ec99402c6f628887@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 ------------[ cut here ]------------ kernel BUG at fs/f2fs/node.c:1185! invalid opcode: 0000 [#1] SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) Modules linked in: CPU: 1 PID: 4549 Comm: syzkaller704305 Not tainted 4.17.0-rc1+ #10 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:__get_node_page+0xb68/0x16e0 fs/f2fs/node.c:1185 RSP: 0018:ffff8801d960e820 EFLAGS: 00010293 RAX: ffff8801d88205c0 RBX: 0000000000000003 RCX: ffffffff82f6cc06 RDX: 0000000000000000 RSI: ffffffff82f6d5e8 RDI: 0000000000000004 RBP: ffff8801d960ec30 R08: ffff8801d88205c0 R09: ffffed003b5e46c2 R10: 0000000000000003 R11: 0000000000000003 R12: ffff8801a86e00c0 R13: 0000000000000001 R14: ffff8801a86e0530 R15: ffff8801d9745240 FS: 000000000072c880(0000) GS:ffff8801daf00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f3d403209b8 CR3: 00000001d8f3f000 CR4: 00000000001406e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: get_node_page fs/f2fs/node.c:1237 [inline] truncate_xattr_node+0x152/0x2e0 fs/f2fs/node.c:1014 remove_inode_page+0x200/0xaf0 fs/f2fs/node.c:1039 f2fs_evict_inode+0xe86/0x1710 fs/f2fs/inode.c:547 evict+0x4a6/0x960 fs/inode.c:557 iput_final fs/inode.c:1519 [inline] iput+0x62d/0xa80 fs/inode.c:1545 f2fs_fill_super+0x5f4e/0x7bf0 fs/f2fs/super.c:2849 mount_bdev+0x30c/0x3e0 fs/super.c:1164 f2fs_mount+0x34/0x40 fs/f2fs/super.c:3020 mount_fs+0xae/0x328 fs/super.c:1267 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:2518 [inline] do_mount+0x564/0x3070 fs/namespace.c:2848 ksys_mount+0x12d/0x140 fs/namespace.c:3064 __do_sys_mount fs/namespace.c:3078 [inline] __se_sys_mount fs/namespace.c:3075 [inline] __x64_sys_mount+0xbe/0x150 fs/namespace.c:3075 do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x443dea RSP: 002b:00007ffcc7882368 EFLAGS: 00000297 ORIG_RAX: 00000000000000a5 RAX: ffffffffffffffda RBX: 0000000020000c00 RCX: 0000000000443dea RDX: 0000000020000000 RSI: 0000000020000100 RDI: 00007ffcc7882370 RBP: 0000000000000003 R08: 0000000020016a00 R09: 000000000000000a R10: 0000000000000000 R11: 0000000000000297 R12: 0000000000000004 R13: 0000000000402ce0 R14: 0000000000000000 R15: 0000000000000000 RIP: __get_node_page+0xb68/0x16e0 fs/f2fs/node.c:1185 RSP: ffff8801d960e820 ---[ end trace 4edbeb71f002bb76 ]--- Reported-and-tested-by: syzbot+d154ec99402c6f628887@syzkaller.appspotmail.com Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-04-24 05:02:31 +00:00
}
return 0;
}
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct sysinfo val;
unsigned long avail_ram;
unsigned long mem_size = 0;
bool res = false;
si_meminfo(&val);
/* only uses low memory */
avail_ram = val.totalram - val.totalhigh;
/*
* give 25%, 25%, 50%, 50%, 50% memory for each components respectively
*/
if (type == FREE_NIDS) {
mem_size = (nm_i->nid_cnt[FREE_NID] *
sizeof(struct free_nid)) >> PAGE_SHIFT;
res = mem_size < ((avail_ram * nm_i->ram_thresh / 100) >> 2);
} else if (type == NAT_ENTRIES) {
mem_size = (nm_i->nat_cnt * sizeof(struct nat_entry)) >>
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_SHIFT;
res = mem_size < ((avail_ram * nm_i->ram_thresh / 100) >> 2);
if (excess_cached_nats(sbi))
res = false;
} else if (type == DIRTY_DENTS) {
if (sbi->sb->s_bdi->wb.dirty_exceeded)
return false;
mem_size = get_pages(sbi, F2FS_DIRTY_DENTS);
res = mem_size < ((avail_ram * nm_i->ram_thresh / 100) >> 1);
} else if (type == INO_ENTRIES) {
int i;
for (i = 0; i < MAX_INO_ENTRY; i++)
mem_size += sbi->im[i].ino_num *
sizeof(struct ino_entry);
mem_size >>= PAGE_SHIFT;
res = mem_size < ((avail_ram * nm_i->ram_thresh / 100) >> 1);
} else if (type == EXTENT_CACHE) {
mem_size = (atomic_read(&sbi->total_ext_tree) *
sizeof(struct extent_tree) +
atomic_read(&sbi->total_ext_node) *
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
sizeof(struct extent_node)) >> PAGE_SHIFT;
res = mem_size < ((avail_ram * nm_i->ram_thresh / 100) >> 1);
} else if (type == INMEM_PAGES) {
/* it allows 20% / total_ram for inmemory pages */
mem_size = get_pages(sbi, F2FS_INMEM_PAGES);
res = mem_size < (val.totalram / 5);
} else {
if (!sbi->sb->s_bdi->wb.dirty_exceeded)
return true;
}
return res;
}
static void clear_node_page_dirty(struct page *page)
{
if (PageDirty(page)) {
f2fs_clear_page_cache_dirty_tag(page);
clear_page_dirty_for_io(page);
dec_page_count(F2FS_P_SB(page), F2FS_DIRTY_NODES);
}
ClearPageUptodate(page);
}
static struct page *get_current_nat_page(struct f2fs_sb_info *sbi, nid_t nid)
{
return f2fs_get_meta_page_nofail(sbi, current_nat_addr(sbi, nid));
}
static struct page *get_next_nat_page(struct f2fs_sb_info *sbi, nid_t nid)
{
struct page *src_page;
struct page *dst_page;
pgoff_t dst_off;
void *src_addr;
void *dst_addr;
struct f2fs_nm_info *nm_i = NM_I(sbi);
dst_off = next_nat_addr(sbi, current_nat_addr(sbi, nid));
/* get current nat block page with lock */
src_page = get_current_nat_page(sbi, nid);
if (IS_ERR(src_page))
return src_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
dst_page = f2fs_grab_meta_page(sbi, dst_off);
f2fs_bug_on(sbi, PageDirty(src_page));
src_addr = page_address(src_page);
dst_addr = page_address(dst_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
memcpy(dst_addr, src_addr, PAGE_SIZE);
set_page_dirty(dst_page);
f2fs_put_page(src_page, 1);
set_to_next_nat(nm_i, nid);
return dst_page;
}
static struct nat_entry *__alloc_nat_entry(nid_t nid, bool no_fail)
{
struct nat_entry *new;
if (no_fail)
new = f2fs_kmem_cache_alloc(nat_entry_slab, GFP_F2FS_ZERO);
else
new = kmem_cache_alloc(nat_entry_slab, GFP_F2FS_ZERO);
if (new) {
nat_set_nid(new, nid);
nat_reset_flag(new);
}
return new;
}
static void __free_nat_entry(struct nat_entry *e)
{
kmem_cache_free(nat_entry_slab, e);
}
/* must be locked by nat_tree_lock */
static struct nat_entry *__init_nat_entry(struct f2fs_nm_info *nm_i,
struct nat_entry *ne, struct f2fs_nat_entry *raw_ne, bool no_fail)
{
if (no_fail)
f2fs_radix_tree_insert(&nm_i->nat_root, nat_get_nid(ne), ne);
else if (radix_tree_insert(&nm_i->nat_root, nat_get_nid(ne), ne))
return NULL;
if (raw_ne)
node_info_from_raw_nat(&ne->ni, raw_ne);
spin_lock(&nm_i->nat_list_lock);
list_add_tail(&ne->list, &nm_i->nat_entries);
spin_unlock(&nm_i->nat_list_lock);
nm_i->nat_cnt++;
return ne;
}
static struct nat_entry *__lookup_nat_cache(struct f2fs_nm_info *nm_i, nid_t n)
{
struct nat_entry *ne;
ne = radix_tree_lookup(&nm_i->nat_root, n);
/* for recent accessed nat entry, move it to tail of lru list */
if (ne && !get_nat_flag(ne, IS_DIRTY)) {
spin_lock(&nm_i->nat_list_lock);
if (!list_empty(&ne->list))
list_move_tail(&ne->list, &nm_i->nat_entries);
spin_unlock(&nm_i->nat_list_lock);
}
return ne;
}
static unsigned int __gang_lookup_nat_cache(struct f2fs_nm_info *nm_i,
nid_t start, unsigned int nr, struct nat_entry **ep)
{
return radix_tree_gang_lookup(&nm_i->nat_root, (void **)ep, start, nr);
}
static void __del_from_nat_cache(struct f2fs_nm_info *nm_i, struct nat_entry *e)
{
radix_tree_delete(&nm_i->nat_root, nat_get_nid(e));
nm_i->nat_cnt--;
__free_nat_entry(e);
}
static struct nat_entry_set *__grab_nat_entry_set(struct f2fs_nm_info *nm_i,
struct nat_entry *ne)
{
nid_t set = NAT_BLOCK_OFFSET(ne->ni.nid);
struct nat_entry_set *head;
head = radix_tree_lookup(&nm_i->nat_set_root, set);
if (!head) {
head = f2fs_kmem_cache_alloc(nat_entry_set_slab, GFP_NOFS);
INIT_LIST_HEAD(&head->entry_list);
INIT_LIST_HEAD(&head->set_list);
head->set = set;
head->entry_cnt = 0;
f2fs_radix_tree_insert(&nm_i->nat_set_root, set, head);
}
return head;
}
static void __set_nat_cache_dirty(struct f2fs_nm_info *nm_i,
struct nat_entry *ne)
{
struct nat_entry_set *head;
bool new_ne = nat_get_blkaddr(ne) == NEW_ADDR;
if (!new_ne)
head = __grab_nat_entry_set(nm_i, ne);
/*
* update entry_cnt in below condition:
* 1. update NEW_ADDR to valid block address;
* 2. update old block address to new one;
*/
if (!new_ne && (get_nat_flag(ne, IS_PREALLOC) ||
!get_nat_flag(ne, IS_DIRTY)))
head->entry_cnt++;
set_nat_flag(ne, IS_PREALLOC, new_ne);
if (get_nat_flag(ne, IS_DIRTY))
goto refresh_list;
nm_i->dirty_nat_cnt++;
set_nat_flag(ne, IS_DIRTY, true);
refresh_list:
spin_lock(&nm_i->nat_list_lock);
if (new_ne)
list_del_init(&ne->list);
else
list_move_tail(&ne->list, &head->entry_list);
spin_unlock(&nm_i->nat_list_lock);
}
static void __clear_nat_cache_dirty(struct f2fs_nm_info *nm_i,
struct nat_entry_set *set, struct nat_entry *ne)
{
spin_lock(&nm_i->nat_list_lock);
list_move_tail(&ne->list, &nm_i->nat_entries);
spin_unlock(&nm_i->nat_list_lock);
set_nat_flag(ne, IS_DIRTY, false);
set->entry_cnt--;
nm_i->dirty_nat_cnt--;
}
static unsigned int __gang_lookup_nat_set(struct f2fs_nm_info *nm_i,
nid_t start, unsigned int nr, struct nat_entry_set **ep)
{
return radix_tree_gang_lookup(&nm_i->nat_set_root, (void **)ep,
start, nr);
}
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page)
{
return NODE_MAPPING(sbi) == page->mapping &&
IS_DNODE(page) && is_cold_node(page);
}
void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi)
{
spin_lock_init(&sbi->fsync_node_lock);
INIT_LIST_HEAD(&sbi->fsync_node_list);
sbi->fsync_seg_id = 0;
sbi->fsync_node_num = 0;
}
static unsigned int f2fs_add_fsync_node_entry(struct f2fs_sb_info *sbi,
struct page *page)
{
struct fsync_node_entry *fn;
unsigned long flags;
unsigned int seq_id;
fn = f2fs_kmem_cache_alloc(fsync_node_entry_slab, GFP_NOFS);
get_page(page);
fn->page = page;
INIT_LIST_HEAD(&fn->list);
spin_lock_irqsave(&sbi->fsync_node_lock, flags);
list_add_tail(&fn->list, &sbi->fsync_node_list);
fn->seq_id = sbi->fsync_seg_id++;
seq_id = fn->seq_id;
sbi->fsync_node_num++;
spin_unlock_irqrestore(&sbi->fsync_node_lock, flags);
return seq_id;
}
void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page)
{
struct fsync_node_entry *fn;
unsigned long flags;
spin_lock_irqsave(&sbi->fsync_node_lock, flags);
list_for_each_entry(fn, &sbi->fsync_node_list, list) {
if (fn->page == page) {
list_del(&fn->list);
sbi->fsync_node_num--;
spin_unlock_irqrestore(&sbi->fsync_node_lock, flags);
kmem_cache_free(fsync_node_entry_slab, fn);
put_page(page);
return;
}
}
spin_unlock_irqrestore(&sbi->fsync_node_lock, flags);
f2fs_bug_on(sbi, 1);
}
void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi)
{
unsigned long flags;
spin_lock_irqsave(&sbi->fsync_node_lock, flags);
sbi->fsync_seg_id = 0;
spin_unlock_irqrestore(&sbi->fsync_node_lock, flags);
}
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_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct nat_entry *e;
bool need = false;
down_read(&nm_i->nat_tree_lock);
e = __lookup_nat_cache(nm_i, nid);
if (e) {
if (!get_nat_flag(e, IS_CHECKPOINTED) &&
!get_nat_flag(e, HAS_FSYNCED_INODE))
need = true;
}
up_read(&nm_i->nat_tree_lock);
return need;
}
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_node(struct f2fs_sb_info *sbi, nid_t nid)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct nat_entry *e;
bool is_cp = true;
down_read(&nm_i->nat_tree_lock);
e = __lookup_nat_cache(nm_i, nid);
if (e && !get_nat_flag(e, IS_CHECKPOINTED))
is_cp = false;
up_read(&nm_i->nat_tree_lock);
return is_cp;
}
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_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct nat_entry *e;
f2fs: fix conditions to remain recovery information in f2fs_sync_file This patch revisited whole the recovery information during the f2fs_sync_file. In this patch, there are three information to make a decision. a) IS_CHECKPOINTED, /* is it checkpointed before? */ b) HAS_FSYNCED_INODE, /* is the inode fsynced before? */ c) HAS_LAST_FSYNC, /* has the latest node fsync mark? */ And, the scenarios for our rule are based on: [Term] F: fsync_mark, D: dentry_mark 1. inode(x) | CP | inode(x) | dnode(F) 2. inode(x) | CP | inode(F) | dnode(F) 3. inode(x) | CP | dnode(F) | inode(x) | inode(F) 4. inode(x) | CP | dnode(F) | inode(F) 5. CP | inode(x) | dnode(F) | inode(DF) 6. CP | inode(DF) | dnode(F) 7. CP | dnode(F) | inode(DF) 8. CP | dnode(F) | inode(x) | inode(DF) For example, #3, the three conditions should be changed as follows. inode(x) | CP | dnode(F) | inode(x) | inode(F) a) x o o o o b) x x x x o c) x o o x o If f2fs_sync_file stops ------^, it should write inode(F) --------------^ So, the need_inode_block_update should return true, since c) get_nat_flag(e, HAS_LAST_FSYNC), is false. For example, #8, CP | alloc | dnode(F) | inode(x) | inode(DF) a) o x x x x b) x x x o c) o o x o If f2fs_sync_file stops -------^, it should write inode(DF) --------------^ Note that, the roll-forward policy should follow this rule, which means, if there are any missing blocks, we doesn't need to recover that inode. Signed-off-by: Huang Ying <ying.huang@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-15 21:50:48 +00:00
bool need_update = true;
down_read(&nm_i->nat_tree_lock);
f2fs: fix conditions to remain recovery information in f2fs_sync_file This patch revisited whole the recovery information during the f2fs_sync_file. In this patch, there are three information to make a decision. a) IS_CHECKPOINTED, /* is it checkpointed before? */ b) HAS_FSYNCED_INODE, /* is the inode fsynced before? */ c) HAS_LAST_FSYNC, /* has the latest node fsync mark? */ And, the scenarios for our rule are based on: [Term] F: fsync_mark, D: dentry_mark 1. inode(x) | CP | inode(x) | dnode(F) 2. inode(x) | CP | inode(F) | dnode(F) 3. inode(x) | CP | dnode(F) | inode(x) | inode(F) 4. inode(x) | CP | dnode(F) | inode(F) 5. CP | inode(x) | dnode(F) | inode(DF) 6. CP | inode(DF) | dnode(F) 7. CP | dnode(F) | inode(DF) 8. CP | dnode(F) | inode(x) | inode(DF) For example, #3, the three conditions should be changed as follows. inode(x) | CP | dnode(F) | inode(x) | inode(F) a) x o o o o b) x x x x o c) x o o x o If f2fs_sync_file stops ------^, it should write inode(F) --------------^ So, the need_inode_block_update should return true, since c) get_nat_flag(e, HAS_LAST_FSYNC), is false. For example, #8, CP | alloc | dnode(F) | inode(x) | inode(DF) a) o x x x x b) x x x o c) o o x o If f2fs_sync_file stops -------^, it should write inode(DF) --------------^ Note that, the roll-forward policy should follow this rule, which means, if there are any missing blocks, we doesn't need to recover that inode. Signed-off-by: Huang Ying <ying.huang@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-15 21:50:48 +00:00
e = __lookup_nat_cache(nm_i, ino);
if (e && get_nat_flag(e, HAS_LAST_FSYNC) &&
(get_nat_flag(e, IS_CHECKPOINTED) ||
get_nat_flag(e, HAS_FSYNCED_INODE)))
need_update = false;
up_read(&nm_i->nat_tree_lock);
f2fs: fix conditions to remain recovery information in f2fs_sync_file This patch revisited whole the recovery information during the f2fs_sync_file. In this patch, there are three information to make a decision. a) IS_CHECKPOINTED, /* is it checkpointed before? */ b) HAS_FSYNCED_INODE, /* is the inode fsynced before? */ c) HAS_LAST_FSYNC, /* has the latest node fsync mark? */ And, the scenarios for our rule are based on: [Term] F: fsync_mark, D: dentry_mark 1. inode(x) | CP | inode(x) | dnode(F) 2. inode(x) | CP | inode(F) | dnode(F) 3. inode(x) | CP | dnode(F) | inode(x) | inode(F) 4. inode(x) | CP | dnode(F) | inode(F) 5. CP | inode(x) | dnode(F) | inode(DF) 6. CP | inode(DF) | dnode(F) 7. CP | dnode(F) | inode(DF) 8. CP | dnode(F) | inode(x) | inode(DF) For example, #3, the three conditions should be changed as follows. inode(x) | CP | dnode(F) | inode(x) | inode(F) a) x o o o o b) x x x x o c) x o o x o If f2fs_sync_file stops ------^, it should write inode(F) --------------^ So, the need_inode_block_update should return true, since c) get_nat_flag(e, HAS_LAST_FSYNC), is false. For example, #8, CP | alloc | dnode(F) | inode(x) | inode(DF) a) o x x x x b) x x x o c) o o x o If f2fs_sync_file stops -------^, it should write inode(DF) --------------^ Note that, the roll-forward policy should follow this rule, which means, if there are any missing blocks, we doesn't need to recover that inode. Signed-off-by: Huang Ying <ying.huang@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-15 21:50:48 +00:00
return need_update;
}
/* must be locked by nat_tree_lock */
static void cache_nat_entry(struct f2fs_sb_info *sbi, nid_t nid,
struct f2fs_nat_entry *ne)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct nat_entry *new, *e;
new = __alloc_nat_entry(nid, false);
if (!new)
return;
down_write(&nm_i->nat_tree_lock);
e = __lookup_nat_cache(nm_i, nid);
if (!e)
e = __init_nat_entry(nm_i, new, ne, false);
else
f2fs_bug_on(sbi, nat_get_ino(e) != le32_to_cpu(ne->ino) ||
nat_get_blkaddr(e) !=
le32_to_cpu(ne->block_addr) ||
nat_get_version(e) != ne->version);
up_write(&nm_i->nat_tree_lock);
if (e != new)
__free_nat_entry(new);
}
static void set_node_addr(struct f2fs_sb_info *sbi, struct node_info *ni,
block_t new_blkaddr, bool fsync_done)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct nat_entry *e;
struct nat_entry *new = __alloc_nat_entry(ni->nid, true);
down_write(&nm_i->nat_tree_lock);
e = __lookup_nat_cache(nm_i, ni->nid);
if (!e) {
e = __init_nat_entry(nm_i, new, NULL, true);
copy_node_info(&e->ni, ni);
f2fs_bug_on(sbi, ni->blk_addr == NEW_ADDR);
} else if (new_blkaddr == NEW_ADDR) {
/*
* when nid is reallocated,
* previous nat entry can be remained in nat cache.
* So, reinitialize it with new information.
*/
copy_node_info(&e->ni, ni);
f2fs_bug_on(sbi, ni->blk_addr != NULL_ADDR);
}
/* let's free early to reduce memory consumption */
if (e != new)
__free_nat_entry(new);
/* sanity check */
f2fs_bug_on(sbi, nat_get_blkaddr(e) != ni->blk_addr);
f2fs_bug_on(sbi, nat_get_blkaddr(e) == NULL_ADDR &&
new_blkaddr == NULL_ADDR);
f2fs_bug_on(sbi, nat_get_blkaddr(e) == NEW_ADDR &&
new_blkaddr == NEW_ADDR);
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
f2fs_bug_on(sbi, __is_valid_data_blkaddr(nat_get_blkaddr(e)) &&
new_blkaddr == NEW_ADDR);
/* increment version no as node is removed */
if (nat_get_blkaddr(e) != NEW_ADDR && new_blkaddr == NULL_ADDR) {
unsigned char version = nat_get_version(e);
nat_set_version(e, inc_node_version(version));
}
/* change address */
nat_set_blkaddr(e, new_blkaddr);
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(new_blkaddr))
f2fs: fix conditions to remain recovery information in f2fs_sync_file This patch revisited whole the recovery information during the f2fs_sync_file. In this patch, there are three information to make a decision. a) IS_CHECKPOINTED, /* is it checkpointed before? */ b) HAS_FSYNCED_INODE, /* is the inode fsynced before? */ c) HAS_LAST_FSYNC, /* has the latest node fsync mark? */ And, the scenarios for our rule are based on: [Term] F: fsync_mark, D: dentry_mark 1. inode(x) | CP | inode(x) | dnode(F) 2. inode(x) | CP | inode(F) | dnode(F) 3. inode(x) | CP | dnode(F) | inode(x) | inode(F) 4. inode(x) | CP | dnode(F) | inode(F) 5. CP | inode(x) | dnode(F) | inode(DF) 6. CP | inode(DF) | dnode(F) 7. CP | dnode(F) | inode(DF) 8. CP | dnode(F) | inode(x) | inode(DF) For example, #3, the three conditions should be changed as follows. inode(x) | CP | dnode(F) | inode(x) | inode(F) a) x o o o o b) x x x x o c) x o o x o If f2fs_sync_file stops ------^, it should write inode(F) --------------^ So, the need_inode_block_update should return true, since c) get_nat_flag(e, HAS_LAST_FSYNC), is false. For example, #8, CP | alloc | dnode(F) | inode(x) | inode(DF) a) o x x x x b) x x x o c) o o x o If f2fs_sync_file stops -------^, it should write inode(DF) --------------^ Note that, the roll-forward policy should follow this rule, which means, if there are any missing blocks, we doesn't need to recover that inode. Signed-off-by: Huang Ying <ying.huang@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-15 21:50:48 +00:00
set_nat_flag(e, IS_CHECKPOINTED, false);
__set_nat_cache_dirty(nm_i, e);
/* update fsync_mark if its inode nat entry is still alive */
if (ni->nid != ni->ino)
e = __lookup_nat_cache(nm_i, ni->ino);
f2fs: fix conditions to remain recovery information in f2fs_sync_file This patch revisited whole the recovery information during the f2fs_sync_file. In this patch, there are three information to make a decision. a) IS_CHECKPOINTED, /* is it checkpointed before? */ b) HAS_FSYNCED_INODE, /* is the inode fsynced before? */ c) HAS_LAST_FSYNC, /* has the latest node fsync mark? */ And, the scenarios for our rule are based on: [Term] F: fsync_mark, D: dentry_mark 1. inode(x) | CP | inode(x) | dnode(F) 2. inode(x) | CP | inode(F) | dnode(F) 3. inode(x) | CP | dnode(F) | inode(x) | inode(F) 4. inode(x) | CP | dnode(F) | inode(F) 5. CP | inode(x) | dnode(F) | inode(DF) 6. CP | inode(DF) | dnode(F) 7. CP | dnode(F) | inode(DF) 8. CP | dnode(F) | inode(x) | inode(DF) For example, #3, the three conditions should be changed as follows. inode(x) | CP | dnode(F) | inode(x) | inode(F) a) x o o o o b) x x x x o c) x o o x o If f2fs_sync_file stops ------^, it should write inode(F) --------------^ So, the need_inode_block_update should return true, since c) get_nat_flag(e, HAS_LAST_FSYNC), is false. For example, #8, CP | alloc | dnode(F) | inode(x) | inode(DF) a) o x x x x b) x x x o c) o o x o If f2fs_sync_file stops -------^, it should write inode(DF) --------------^ Note that, the roll-forward policy should follow this rule, which means, if there are any missing blocks, we doesn't need to recover that inode. Signed-off-by: Huang Ying <ying.huang@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-15 21:50:48 +00:00
if (e) {
if (fsync_done && ni->nid == ni->ino)
set_nat_flag(e, HAS_FSYNCED_INODE, true);
set_nat_flag(e, HAS_LAST_FSYNC, fsync_done);
}
up_write(&nm_i->nat_tree_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_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
int nr = nr_shrink;
if (!down_write_trylock(&nm_i->nat_tree_lock))
return 0;
spin_lock(&nm_i->nat_list_lock);
while (nr_shrink) {
struct nat_entry *ne;
if (list_empty(&nm_i->nat_entries))
break;
ne = list_first_entry(&nm_i->nat_entries,
struct nat_entry, list);
list_del(&ne->list);
spin_unlock(&nm_i->nat_list_lock);
__del_from_nat_cache(nm_i, ne);
nr_shrink--;
spin_lock(&nm_i->nat_list_lock);
}
spin_unlock(&nm_i->nat_list_lock);
up_write(&nm_i->nat_tree_lock);
return nr - nr_shrink;
}
/*
* This function always returns success
*/
int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
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 node_info *ni)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA);
struct f2fs_journal *journal = curseg->journal;
nid_t start_nid = START_NID(nid);
struct f2fs_nat_block *nat_blk;
struct page *page = NULL;
struct f2fs_nat_entry ne;
struct nat_entry *e;
pgoff_t index;
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
block_t blkaddr;
int i;
ni->nid = nid;
/* Check nat cache */
down_read(&nm_i->nat_tree_lock);
e = __lookup_nat_cache(nm_i, nid);
if (e) {
ni->ino = nat_get_ino(e);
ni->blk_addr = nat_get_blkaddr(e);
ni->version = nat_get_version(e);
up_read(&nm_i->nat_tree_lock);
return 0;
}
memset(&ne, 0, sizeof(struct f2fs_nat_entry));
/* Check current segment summary */
down_read(&curseg->journal_rwsem);
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
i = f2fs_lookup_journal_in_cursum(journal, NAT_JOURNAL, nid, 0);
if (i >= 0) {
ne = nat_in_journal(journal, i);
node_info_from_raw_nat(ni, &ne);
}
up_read(&curseg->journal_rwsem);
if (i >= 0) {
up_read(&nm_i->nat_tree_lock);
goto cache;
}
/* Fill node_info from nat page */
index = current_nat_addr(sbi, nid);
up_read(&nm_i->nat_tree_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
page = f2fs_get_meta_page(sbi, index);
if (IS_ERR(page))
return PTR_ERR(page);
nat_blk = (struct f2fs_nat_block *)page_address(page);
ne = nat_blk->entries[nid - start_nid];
node_info_from_raw_nat(ni, &ne);
f2fs_put_page(page, 1);
cache:
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
blkaddr = le32_to_cpu(ne.block_addr);
if (__is_valid_data_blkaddr(blkaddr) &&
!f2fs_is_valid_blkaddr(sbi, blkaddr, DATA_GENERIC_ENHANCE))
return -EFAULT;
/* cache nat entry */
cache_nat_entry(sbi, nid, &ne);
return 0;
}
/*
* readahead MAX_RA_NODE number of node pages.
*/
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 void f2fs_ra_node_pages(struct page *parent, int start, int n)
{
struct f2fs_sb_info *sbi = F2FS_P_SB(parent);
struct blk_plug plug;
int i, end;
nid_t nid;
blk_start_plug(&plug);
/* Then, try readahead for siblings of the desired node */
end = start + n;
end = min(end, NIDS_PER_BLOCK);
for (i = start; i < end; i++) {
nid = get_nid(parent, i, 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_ra_node_page(sbi, nid);
}
blk_finish_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
pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs)
{
const long direct_index = ADDRS_PER_INODE(dn->inode);
const long direct_blks = ADDRS_PER_BLOCK(dn->inode);
const long indirect_blks = ADDRS_PER_BLOCK(dn->inode) * NIDS_PER_BLOCK;
unsigned int skipped_unit = ADDRS_PER_BLOCK(dn->inode);
int cur_level = dn->cur_level;
int max_level = dn->max_level;
pgoff_t base = 0;
if (!dn->max_level)
return pgofs + 1;
while (max_level-- > cur_level)
skipped_unit *= NIDS_PER_BLOCK;
switch (dn->max_level) {
case 3:
base += 2 * indirect_blks;
fs: mark expected switch fall-throughs In preparation to enabling -Wimplicit-fallthrough, mark switch cases where we are expecting to fall through. This patch fixes the following warnings: fs/affs/affs.h:124:38: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/configfs/dir.c:1692:11: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/configfs/dir.c:1694:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ceph/file.c:249:3: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext4/hash.c:233:15: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext4/hash.c:246:15: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext2/inode.c:1237:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext2/inode.c:1244:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext4/indirect.c:1182:6: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext4/indirect.c:1188:6: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext4/indirect.c:1432:6: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext4/indirect.c:1440:6: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/f2fs/node.c:618:8: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/f2fs/node.c:620:8: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/btrfs/ref-verify.c:522:15: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/gfs2/bmap.c:711:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/gfs2/bmap.c:722:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/jffs2/fs.c:339:6: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/nfsd/nfs4proc.c:429:12: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ufs/util.h:62:6: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ufs/util.h:43:6: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/fcntl.c:770:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/seq_file.c:319:10: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/libfs.c:148:11: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/libfs.c:150:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/signalfd.c:178:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/locks.c:1473:16: warning: this statement may fall through [-Wimplicit-fallthrough=] Warning level 3 was used: -Wimplicit-fallthrough=3 This patch is part of the ongoing efforts to enabling -Wimplicit-fallthrough. Reviewed-by: Kees Cook <keescook@chromium.org> Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
2019-01-23 08:48:28 +00:00
/* fall through */
case 2:
base += 2 * direct_blks;
fs: mark expected switch fall-throughs In preparation to enabling -Wimplicit-fallthrough, mark switch cases where we are expecting to fall through. This patch fixes the following warnings: fs/affs/affs.h:124:38: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/configfs/dir.c:1692:11: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/configfs/dir.c:1694:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ceph/file.c:249:3: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext4/hash.c:233:15: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext4/hash.c:246:15: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext2/inode.c:1237:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext2/inode.c:1244:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext4/indirect.c:1182:6: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext4/indirect.c:1188:6: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext4/indirect.c:1432:6: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ext4/indirect.c:1440:6: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/f2fs/node.c:618:8: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/f2fs/node.c:620:8: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/btrfs/ref-verify.c:522:15: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/gfs2/bmap.c:711:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/gfs2/bmap.c:722:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/jffs2/fs.c:339:6: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/nfsd/nfs4proc.c:429:12: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ufs/util.h:62:6: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/ufs/util.h:43:6: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/fcntl.c:770:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/seq_file.c:319:10: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/libfs.c:148:11: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/libfs.c:150:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/signalfd.c:178:7: warning: this statement may fall through [-Wimplicit-fallthrough=] fs/locks.c:1473:16: warning: this statement may fall through [-Wimplicit-fallthrough=] Warning level 3 was used: -Wimplicit-fallthrough=3 This patch is part of the ongoing efforts to enabling -Wimplicit-fallthrough. Reviewed-by: Kees Cook <keescook@chromium.org> Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
2019-01-23 08:48:28 +00:00
/* fall through */
case 1:
base += direct_index;
break;
default:
f2fs_bug_on(F2FS_I_SB(dn->inode), 1);
}
return ((pgofs - base) / skipped_unit + 1) * skipped_unit + base;
}
/*
* The maximum depth is four.
* Offset[0] will have raw inode offset.
*/
static int get_node_path(struct inode *inode, long block,
int offset[4], unsigned int noffset[4])
{
const long direct_index = ADDRS_PER_INODE(inode);
const long direct_blks = ADDRS_PER_BLOCK(inode);
const long dptrs_per_blk = NIDS_PER_BLOCK;
const long indirect_blks = ADDRS_PER_BLOCK(inode) * NIDS_PER_BLOCK;
const long dindirect_blks = indirect_blks * NIDS_PER_BLOCK;
int n = 0;
int level = 0;
noffset[0] = 0;
if (block < direct_index) {
offset[n] = block;
goto got;
}
block -= direct_index;
if (block < direct_blks) {
offset[n++] = NODE_DIR1_BLOCK;
noffset[n] = 1;
offset[n] = block;
level = 1;
goto got;
}
block -= direct_blks;
if (block < direct_blks) {
offset[n++] = NODE_DIR2_BLOCK;
noffset[n] = 2;
offset[n] = block;
level = 1;
goto got;
}
block -= direct_blks;
if (block < indirect_blks) {
offset[n++] = NODE_IND1_BLOCK;
noffset[n] = 3;
offset[n++] = block / direct_blks;
noffset[n] = 4 + offset[n - 1];
offset[n] = block % direct_blks;
level = 2;
goto got;
}
block -= indirect_blks;
if (block < indirect_blks) {
offset[n++] = NODE_IND2_BLOCK;
noffset[n] = 4 + dptrs_per_blk;
offset[n++] = block / direct_blks;
noffset[n] = 5 + dptrs_per_blk + offset[n - 1];
offset[n] = block % direct_blks;
level = 2;
goto got;
}
block -= indirect_blks;
if (block < dindirect_blks) {
offset[n++] = NODE_DIND_BLOCK;
noffset[n] = 5 + (dptrs_per_blk * 2);
offset[n++] = block / indirect_blks;
noffset[n] = 6 + (dptrs_per_blk * 2) +
offset[n - 1] * (dptrs_per_blk + 1);
offset[n++] = (block / direct_blks) % dptrs_per_blk;
noffset[n] = 7 + (dptrs_per_blk * 2) +
offset[n - 2] * (dptrs_per_blk + 1) +
offset[n - 1];
offset[n] = block % direct_blks;
level = 3;
goto got;
} else {
return -E2BIG;
}
got:
return level;
}
/*
* Caller should call f2fs_put_dnode(dn).
* Also, it should grab and release a rwsem by calling f2fs_lock_op() and
* f2fs_unlock_op() only if ro is not set RDONLY_NODE.
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
* In the case of RDONLY_NODE, we don't need to care about 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
int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(dn->inode);
struct page *npage[4];
struct page *parent = NULL;
int offset[4];
unsigned int noffset[4];
nid_t nids[4];
int level, i = 0;
int err = 0;
level = get_node_path(dn->inode, index, offset, noffset);
if (level < 0)
return level;
nids[0] = dn->inode->i_ino;
npage[0] = dn->inode_page;
if (!npage[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
npage[0] = f2fs_get_node_page(sbi, nids[0]);
if (IS_ERR(npage[0]))
return PTR_ERR(npage[0]);
}
/* if inline_data is set, should not report any block indices */
if (f2fs_has_inline_data(dn->inode) && index) {
err = -ENOENT;
f2fs_put_page(npage[0], 1);
goto release_out;
}
parent = npage[0];
if (level != 0)
nids[1] = get_nid(parent, offset[0], true);
dn->inode_page = npage[0];
dn->inode_page_locked = true;
/* get indirect or direct nodes */
for (i = 1; i <= level; i++) {
bool done = false;
if (!nids[i] && mode == ALLOC_NODE) {
/* alloc new 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
if (!f2fs_alloc_nid(sbi, &(nids[i]))) {
err = -ENOSPC;
goto release_pages;
}
dn->nid = nids[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
npage[i] = f2fs_new_node_page(dn, noffset[i]);
if (IS_ERR(npage[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
f2fs_alloc_nid_failed(sbi, nids[i]);
err = PTR_ERR(npage[i]);
goto release_pages;
}
set_nid(parent, offset[i - 1], nids[i], i == 1);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
f2fs_alloc_nid_done(sbi, nids[i]);
done = true;
} else if (mode == LOOKUP_NODE_RA && i == level && level > 1) {
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
npage[i] = f2fs_get_node_page_ra(parent, offset[i - 1]);
if (IS_ERR(npage[i])) {
err = PTR_ERR(npage[i]);
goto release_pages;
}
done = true;
}
if (i == 1) {
dn->inode_page_locked = false;
unlock_page(parent);
} else {
f2fs_put_page(parent, 1);
}
if (!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
npage[i] = f2fs_get_node_page(sbi, nids[i]);
if (IS_ERR(npage[i])) {
err = PTR_ERR(npage[i]);
f2fs_put_page(npage[0], 0);
goto release_out;
}
}
if (i < level) {
parent = npage[i];
nids[i + 1] = get_nid(parent, offset[i], false);
}
}
dn->nid = nids[level];
dn->ofs_in_node = offset[level];
dn->node_page = npage[level];
f2fs: enhance on-disk inode structure scalability This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline to indicate that on-disk structure of current inode is extended. In order to extend, we changed the inode structure a bit: Original one: struct f2fs_inode { ... struct f2fs_extent i_ext; __le32 i_addr[DEF_ADDRS_PER_INODE]; __le32 i_nid[DEF_NIDS_PER_INODE]; } Extended one: struct f2fs_inode { ... struct f2fs_extent i_ext; union { struct { __le16 i_extra_isize; __le16 i_padding; __le32 i_extra_end[0]; }; __le32 i_addr[DEF_ADDRS_PER_INODE]; }; __le32 i_nid[DEF_NIDS_PER_INODE]; } Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of i_addr field for storing i_extra_isize and i_padding. with i_extra_isize, we can calculate actual size of reserved space in i_addr, available attribute fields included in total extra attribute fields for current inode can be described as below: +--------------------+ | .i_mode | | ... | | .i_ext | +--------------------+ | .i_extra_isize |-----+ | .i_padding | | | .i_prjid | | | .i_atime_extra | | | .i_ctime_extra | | | .i_mtime_extra |<----+ | .i_inode_cs |<----- store blkaddr/inline from here | .i_xattr_cs | | ... | +--------------------+ | | | block address | | | +--------------------+ | .i_nid | +--------------------+ | node_footer | | (nid, ino, offset) | +--------------------+ Hence, with this patch, we would enhance scalability of f2fs inode for storing more newly added attribute. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-18 16:19:06 +00:00
dn->data_blkaddr = datablock_addr(dn->inode,
dn->node_page, dn->ofs_in_node);
return 0;
release_pages:
f2fs_put_page(parent, 1);
if (i > 1)
f2fs_put_page(npage[0], 0);
release_out:
dn->inode_page = NULL;
dn->node_page = NULL;
if (err == -ENOENT) {
dn->cur_level = i;
dn->max_level = level;
dn->ofs_in_node = offset[level];
}
return err;
}
static int truncate_node(struct dnode_of_data *dn)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(dn->inode);
struct node_info ni;
int err;
pgoff_t index;
err = f2fs_get_node_info(sbi, dn->nid, &ni);
if (err)
return err;
/* Deallocate node address */
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, ni.blk_addr);
dec_valid_node_count(sbi, dn->inode, dn->nid == dn->inode->i_ino);
set_node_addr(sbi, &ni, NULL_ADDR, false);
if (dn->nid == dn->inode->i_ino) {
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
f2fs_remove_orphan_inode(sbi, dn->nid);
dec_valid_inode_count(sbi);
f2fs_inode_synced(dn->inode);
}
clear_node_page_dirty(dn->node_page);
set_sbi_flag(sbi, SBI_IS_DIRTY);
index = dn->node_page->index;
f2fs_put_page(dn->node_page, 1);
invalidate_mapping_pages(NODE_MAPPING(sbi),
index, index);
dn->node_page = NULL;
trace_f2fs_truncate_node(dn->inode, dn->nid, ni.blk_addr);
return 0;
}
static int truncate_dnode(struct dnode_of_data *dn)
{
struct page *page;
int err;
if (dn->nid == 0)
return 1;
/* get direct 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
page = f2fs_get_node_page(F2FS_I_SB(dn->inode), dn->nid);
if (IS_ERR(page) && PTR_ERR(page) == -ENOENT)
return 1;
else if (IS_ERR(page))
return PTR_ERR(page);
/* Make dnode_of_data for parameter */
dn->node_page = page;
dn->ofs_in_node = 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
f2fs_truncate_data_blocks(dn);
err = truncate_node(dn);
if (err)
return err;
return 1;
}
static int truncate_nodes(struct dnode_of_data *dn, unsigned int nofs,
int ofs, int depth)
{
struct dnode_of_data rdn = *dn;
struct page *page;
struct f2fs_node *rn;
nid_t child_nid;
unsigned int child_nofs;
int freed = 0;
int i, ret;
if (dn->nid == 0)
return NIDS_PER_BLOCK + 1;
trace_f2fs_truncate_nodes_enter(dn->inode, dn->nid, dn->data_blkaddr);
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_node_page(F2FS_I_SB(dn->inode), dn->nid);
if (IS_ERR(page)) {
trace_f2fs_truncate_nodes_exit(dn->inode, PTR_ERR(page));
return PTR_ERR(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
f2fs_ra_node_pages(page, ofs, NIDS_PER_BLOCK);
rn = F2FS_NODE(page);
if (depth < 3) {
for (i = ofs; i < NIDS_PER_BLOCK; i++, freed++) {
child_nid = le32_to_cpu(rn->in.nid[i]);
if (child_nid == 0)
continue;
rdn.nid = child_nid;
ret = truncate_dnode(&rdn);
if (ret < 0)
goto out_err;
if (set_nid(page, i, 0, false))
dn->node_changed = true;
}
} else {
child_nofs = nofs + ofs * (NIDS_PER_BLOCK + 1) + 1;
for (i = ofs; i < NIDS_PER_BLOCK; i++) {
child_nid = le32_to_cpu(rn->in.nid[i]);
if (child_nid == 0) {
child_nofs += NIDS_PER_BLOCK + 1;
continue;
}
rdn.nid = child_nid;
ret = truncate_nodes(&rdn, child_nofs, 0, depth - 1);
if (ret == (NIDS_PER_BLOCK + 1)) {
if (set_nid(page, i, 0, false))
dn->node_changed = true;
child_nofs += ret;
} else if (ret < 0 && ret != -ENOENT) {
goto out_err;
}
}
freed = child_nofs;
}
if (!ofs) {
/* remove current indirect node */
dn->node_page = page;
ret = truncate_node(dn);
if (ret)
goto out_err;
freed++;
} else {
f2fs_put_page(page, 1);
}
trace_f2fs_truncate_nodes_exit(dn->inode, freed);
return freed;
out_err:
f2fs_put_page(page, 1);
trace_f2fs_truncate_nodes_exit(dn->inode, ret);
return ret;
}
static int truncate_partial_nodes(struct dnode_of_data *dn,
struct f2fs_inode *ri, int *offset, int depth)
{
struct page *pages[2];
nid_t nid[3];
nid_t child_nid;
int err = 0;
int i;
int idx = depth - 2;
nid[0] = le32_to_cpu(ri->i_nid[offset[0] - NODE_DIR1_BLOCK]);
if (!nid[0])
return 0;
/* get indirect nodes in the path */
f2fs: fix truncate_partial_nodes bug The truncate_partial_nodes puts pages incorrectly in the following two cases. Note that the value for argc 'depth' can only be 2 or 3. Please see truncate_inode_blocks() and truncate_partial_nodes(). 1) An err is occurred in the first 'for' loop When err is occurred with depth = 2, pages[0] is invalid, so this page doesn't need to be put. There is no problem, however, when depth is 3, it doesn't put the pages correctly where pages[0] is valid and pages[1] is invalid. In this case, depth is set to 2 (ref to statemnt depth = i + 1), and then 'goto fail'. In label 'fail', for (i = depth - 3; i >= 0; i--) cannot meet the condition because i = -1, so pages[0] cann't be put. 2) An err happened in the second 'for' loop Now we've got pages[0] with depth = 2, or we've got pages[0] and pages[1] with depth = 3. When an err is detected, we need 'goto fail' to put such the pages. When depth is 2, in label 'fail', for (i = depth - 3; i >= 0; i--) cann't meet the condition because i = -1, so pages[0] cann't be put. When depth is 3, in label 'fail', for (i = depth - 3; i >= 0; i--) can only put pages[0], pages[1] also cann't be put. Note that 'depth' has been changed before first 'goto fail' (ref to statemnt depth = i + 1), so passing this modified 'depth' to the tracepoint, trace_f2fs_truncate_partial_nodes, is also incorrect. Signed-off-by: Shifei Ge <shifei10.ge@samsung.com> [Jaegeuk Kim: modify the description and fix one bug] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-29 07:32:34 +00:00
for (i = 0; i < idx + 1; i++) {
/* reference count'll be increased */
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
pages[i] = f2fs_get_node_page(F2FS_I_SB(dn->inode), nid[i]);
if (IS_ERR(pages[i])) {
err = PTR_ERR(pages[i]);
f2fs: fix truncate_partial_nodes bug The truncate_partial_nodes puts pages incorrectly in the following two cases. Note that the value for argc 'depth' can only be 2 or 3. Please see truncate_inode_blocks() and truncate_partial_nodes(). 1) An err is occurred in the first 'for' loop When err is occurred with depth = 2, pages[0] is invalid, so this page doesn't need to be put. There is no problem, however, when depth is 3, it doesn't put the pages correctly where pages[0] is valid and pages[1] is invalid. In this case, depth is set to 2 (ref to statemnt depth = i + 1), and then 'goto fail'. In label 'fail', for (i = depth - 3; i >= 0; i--) cannot meet the condition because i = -1, so pages[0] cann't be put. 2) An err happened in the second 'for' loop Now we've got pages[0] with depth = 2, or we've got pages[0] and pages[1] with depth = 3. When an err is detected, we need 'goto fail' to put such the pages. When depth is 2, in label 'fail', for (i = depth - 3; i >= 0; i--) cann't meet the condition because i = -1, so pages[0] cann't be put. When depth is 3, in label 'fail', for (i = depth - 3; i >= 0; i--) can only put pages[0], pages[1] also cann't be put. Note that 'depth' has been changed before first 'goto fail' (ref to statemnt depth = i + 1), so passing this modified 'depth' to the tracepoint, trace_f2fs_truncate_partial_nodes, is also incorrect. Signed-off-by: Shifei Ge <shifei10.ge@samsung.com> [Jaegeuk Kim: modify the description and fix one bug] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-29 07:32:34 +00:00
idx = i - 1;
goto fail;
}
nid[i + 1] = get_nid(pages[i], offset[i + 1], 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_ra_node_pages(pages[idx], offset[idx + 1], NIDS_PER_BLOCK);
/* free direct nodes linked to a partial indirect node */
f2fs: fix truncate_partial_nodes bug The truncate_partial_nodes puts pages incorrectly in the following two cases. Note that the value for argc 'depth' can only be 2 or 3. Please see truncate_inode_blocks() and truncate_partial_nodes(). 1) An err is occurred in the first 'for' loop When err is occurred with depth = 2, pages[0] is invalid, so this page doesn't need to be put. There is no problem, however, when depth is 3, it doesn't put the pages correctly where pages[0] is valid and pages[1] is invalid. In this case, depth is set to 2 (ref to statemnt depth = i + 1), and then 'goto fail'. In label 'fail', for (i = depth - 3; i >= 0; i--) cannot meet the condition because i = -1, so pages[0] cann't be put. 2) An err happened in the second 'for' loop Now we've got pages[0] with depth = 2, or we've got pages[0] and pages[1] with depth = 3. When an err is detected, we need 'goto fail' to put such the pages. When depth is 2, in label 'fail', for (i = depth - 3; i >= 0; i--) cann't meet the condition because i = -1, so pages[0] cann't be put. When depth is 3, in label 'fail', for (i = depth - 3; i >= 0; i--) can only put pages[0], pages[1] also cann't be put. Note that 'depth' has been changed before first 'goto fail' (ref to statemnt depth = i + 1), so passing this modified 'depth' to the tracepoint, trace_f2fs_truncate_partial_nodes, is also incorrect. Signed-off-by: Shifei Ge <shifei10.ge@samsung.com> [Jaegeuk Kim: modify the description and fix one bug] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-29 07:32:34 +00:00
for (i = offset[idx + 1]; i < NIDS_PER_BLOCK; i++) {
child_nid = get_nid(pages[idx], i, false);
if (!child_nid)
continue;
dn->nid = child_nid;
err = truncate_dnode(dn);
if (err < 0)
goto fail;
if (set_nid(pages[idx], i, 0, false))
dn->node_changed = true;
}
f2fs: fix truncate_partial_nodes bug The truncate_partial_nodes puts pages incorrectly in the following two cases. Note that the value for argc 'depth' can only be 2 or 3. Please see truncate_inode_blocks() and truncate_partial_nodes(). 1) An err is occurred in the first 'for' loop When err is occurred with depth = 2, pages[0] is invalid, so this page doesn't need to be put. There is no problem, however, when depth is 3, it doesn't put the pages correctly where pages[0] is valid and pages[1] is invalid. In this case, depth is set to 2 (ref to statemnt depth = i + 1), and then 'goto fail'. In label 'fail', for (i = depth - 3; i >= 0; i--) cannot meet the condition because i = -1, so pages[0] cann't be put. 2) An err happened in the second 'for' loop Now we've got pages[0] with depth = 2, or we've got pages[0] and pages[1] with depth = 3. When an err is detected, we need 'goto fail' to put such the pages. When depth is 2, in label 'fail', for (i = depth - 3; i >= 0; i--) cann't meet the condition because i = -1, so pages[0] cann't be put. When depth is 3, in label 'fail', for (i = depth - 3; i >= 0; i--) can only put pages[0], pages[1] also cann't be put. Note that 'depth' has been changed before first 'goto fail' (ref to statemnt depth = i + 1), so passing this modified 'depth' to the tracepoint, trace_f2fs_truncate_partial_nodes, is also incorrect. Signed-off-by: Shifei Ge <shifei10.ge@samsung.com> [Jaegeuk Kim: modify the description and fix one bug] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-29 07:32:34 +00:00
if (offset[idx + 1] == 0) {
dn->node_page = pages[idx];
dn->nid = nid[idx];
err = truncate_node(dn);
if (err)
goto fail;
} else {
f2fs_put_page(pages[idx], 1);
}
offset[idx]++;
f2fs: fix truncate_partial_nodes bug The truncate_partial_nodes puts pages incorrectly in the following two cases. Note that the value for argc 'depth' can only be 2 or 3. Please see truncate_inode_blocks() and truncate_partial_nodes(). 1) An err is occurred in the first 'for' loop When err is occurred with depth = 2, pages[0] is invalid, so this page doesn't need to be put. There is no problem, however, when depth is 3, it doesn't put the pages correctly where pages[0] is valid and pages[1] is invalid. In this case, depth is set to 2 (ref to statemnt depth = i + 1), and then 'goto fail'. In label 'fail', for (i = depth - 3; i >= 0; i--) cannot meet the condition because i = -1, so pages[0] cann't be put. 2) An err happened in the second 'for' loop Now we've got pages[0] with depth = 2, or we've got pages[0] and pages[1] with depth = 3. When an err is detected, we need 'goto fail' to put such the pages. When depth is 2, in label 'fail', for (i = depth - 3; i >= 0; i--) cann't meet the condition because i = -1, so pages[0] cann't be put. When depth is 3, in label 'fail', for (i = depth - 3; i >= 0; i--) can only put pages[0], pages[1] also cann't be put. Note that 'depth' has been changed before first 'goto fail' (ref to statemnt depth = i + 1), so passing this modified 'depth' to the tracepoint, trace_f2fs_truncate_partial_nodes, is also incorrect. Signed-off-by: Shifei Ge <shifei10.ge@samsung.com> [Jaegeuk Kim: modify the description and fix one bug] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-29 07:32:34 +00:00
offset[idx + 1] = 0;
idx--;
fail:
f2fs: fix truncate_partial_nodes bug The truncate_partial_nodes puts pages incorrectly in the following two cases. Note that the value for argc 'depth' can only be 2 or 3. Please see truncate_inode_blocks() and truncate_partial_nodes(). 1) An err is occurred in the first 'for' loop When err is occurred with depth = 2, pages[0] is invalid, so this page doesn't need to be put. There is no problem, however, when depth is 3, it doesn't put the pages correctly where pages[0] is valid and pages[1] is invalid. In this case, depth is set to 2 (ref to statemnt depth = i + 1), and then 'goto fail'. In label 'fail', for (i = depth - 3; i >= 0; i--) cannot meet the condition because i = -1, so pages[0] cann't be put. 2) An err happened in the second 'for' loop Now we've got pages[0] with depth = 2, or we've got pages[0] and pages[1] with depth = 3. When an err is detected, we need 'goto fail' to put such the pages. When depth is 2, in label 'fail', for (i = depth - 3; i >= 0; i--) cann't meet the condition because i = -1, so pages[0] cann't be put. When depth is 3, in label 'fail', for (i = depth - 3; i >= 0; i--) can only put pages[0], pages[1] also cann't be put. Note that 'depth' has been changed before first 'goto fail' (ref to statemnt depth = i + 1), so passing this modified 'depth' to the tracepoint, trace_f2fs_truncate_partial_nodes, is also incorrect. Signed-off-by: Shifei Ge <shifei10.ge@samsung.com> [Jaegeuk Kim: modify the description and fix one bug] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-29 07:32:34 +00:00
for (i = idx; i >= 0; i--)
f2fs_put_page(pages[i], 1);
trace_f2fs_truncate_partial_nodes(dn->inode, nid, depth, err);
return err;
}
/*
* All the block addresses of data and nodes should be nullified.
*/
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_truncate_inode_blocks(struct inode *inode, pgoff_t from)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
int err = 0, cont = 1;
int level, offset[4], noffset[4];
unsigned int nofs = 0;
struct f2fs_inode *ri;
struct dnode_of_data dn;
struct page *page;
trace_f2fs_truncate_inode_blocks_enter(inode, from);
level = get_node_path(inode, from, offset, noffset);
if (level < 0)
return level;
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_node_page(sbi, inode->i_ino);
if (IS_ERR(page)) {
trace_f2fs_truncate_inode_blocks_exit(inode, PTR_ERR(page));
return PTR_ERR(page);
}
set_new_dnode(&dn, inode, page, NULL, 0);
unlock_page(page);
ri = F2FS_INODE(page);
switch (level) {
case 0:
case 1:
nofs = noffset[1];
break;
case 2:
nofs = noffset[1];
if (!offset[level - 1])
goto skip_partial;
err = truncate_partial_nodes(&dn, ri, offset, level);
if (err < 0 && err != -ENOENT)
goto fail;
nofs += 1 + NIDS_PER_BLOCK;
break;
case 3:
nofs = 5 + 2 * NIDS_PER_BLOCK;
if (!offset[level - 1])
goto skip_partial;
err = truncate_partial_nodes(&dn, ri, offset, level);
if (err < 0 && err != -ENOENT)
goto fail;
break;
default:
BUG();
}
skip_partial:
while (cont) {
dn.nid = le32_to_cpu(ri->i_nid[offset[0] - NODE_DIR1_BLOCK]);
switch (offset[0]) {
case NODE_DIR1_BLOCK:
case NODE_DIR2_BLOCK:
err = truncate_dnode(&dn);
break;
case NODE_IND1_BLOCK:
case NODE_IND2_BLOCK:
err = truncate_nodes(&dn, nofs, offset[1], 2);
break;
case NODE_DIND_BLOCK:
err = truncate_nodes(&dn, nofs, offset[1], 3);
cont = 0;
break;
default:
BUG();
}
if (err < 0 && err != -ENOENT)
goto fail;
if (offset[1] == 0 &&
ri->i_nid[offset[0] - NODE_DIR1_BLOCK]) {
lock_page(page);
BUG_ON(page->mapping != NODE_MAPPING(sbi));
f2fs_wait_on_page_writeback(page, NODE, true, true);
ri->i_nid[offset[0] - NODE_DIR1_BLOCK] = 0;
set_page_dirty(page);
unlock_page(page);
}
offset[1] = 0;
offset[0]++;
nofs += err;
}
fail:
f2fs_put_page(page, 0);
trace_f2fs_truncate_inode_blocks_exit(inode, err);
return err > 0 ? 0 : err;
}
/* caller must lock inode 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
int f2fs_truncate_xattr_node(struct inode *inode)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
nid_t nid = F2FS_I(inode)->i_xattr_nid;
struct dnode_of_data dn;
struct page *npage;
int err;
if (!nid)
return 0;
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
npage = f2fs_get_node_page(sbi, nid);
if (IS_ERR(npage))
return PTR_ERR(npage);
set_new_dnode(&dn, inode, NULL, npage, nid);
err = truncate_node(&dn);
if (err) {
f2fs_put_page(npage, 1);
return err;
}
f2fs_i_xnid_write(inode, 0);
return 0;
}
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
/*
* Caller should grab and release a rwsem by calling f2fs_lock_op() and
* f2fs_unlock_op().
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
*/
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_remove_inode_page(struct inode *inode)
{
struct dnode_of_data dn;
int err;
set_new_dnode(&dn, inode, NULL, NULL, inode->i_ino);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
err = f2fs_get_dnode_of_data(&dn, 0, LOOKUP_NODE);
if (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
err = f2fs_truncate_xattr_node(inode);
if (err) {
f2fs_put_dnode(&dn);
return err;
}
/* remove potential inline_data blocks */
if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
S_ISLNK(inode->i_mode))
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_truncate_data_blocks_range(&dn, 1);
/* 0 is possible, after f2fs_new_inode() has failed */
if (unlikely(f2fs_cp_error(F2FS_I_SB(inode)))) {
f2fs_put_dnode(&dn);
return -EIO;
}
if (unlikely(inode->i_blocks != 0 && inode->i_blocks != 8)) {
f2fs_warn(F2FS_I_SB(inode), "Inconsistent i_blocks, ino:%lu, iblocks:%llu",
inode->i_ino, (unsigned long long)inode->i_blocks);
set_sbi_flag(F2FS_I_SB(inode), SBI_NEED_FSCK);
}
/* will put inode & node pages */
err = truncate_node(&dn);
if (err) {
f2fs_put_dnode(&dn);
return err;
}
return 0;
}
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
struct page *f2fs_new_inode_page(struct inode *inode)
{
struct dnode_of_data dn;
/* allocate inode page for new inode */
set_new_dnode(&dn, inode, NULL, NULL, inode->i_ino);
/* caller should f2fs_put_page(page, 1); */
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
return f2fs_new_node_page(&dn, 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
struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(dn->inode);
struct node_info new_ni;
struct page *page;
int err;
if (unlikely(is_inode_flag_set(dn->inode, FI_NO_ALLOC)))
return ERR_PTR(-EPERM);
page = f2fs_grab_cache_page(NODE_MAPPING(sbi), dn->nid, false);
if (!page)
return ERR_PTR(-ENOMEM);
if (unlikely((err = inc_valid_node_count(sbi, dn->inode, !ofs))))
goto fail;
#ifdef CONFIG_F2FS_CHECK_FS
err = f2fs_get_node_info(sbi, dn->nid, &new_ni);
if (err) {
dec_valid_node_count(sbi, dn->inode, !ofs);
goto fail;
}
f2fs_bug_on(sbi, new_ni.blk_addr != NULL_ADDR);
#endif
new_ni.nid = dn->nid;
new_ni.ino = dn->inode->i_ino;
new_ni.blk_addr = NULL_ADDR;
new_ni.flag = 0;
new_ni.version = 0;
set_node_addr(sbi, &new_ni, NEW_ADDR, false);
f2fs_wait_on_page_writeback(page, NODE, true, true);
fill_node_footer(page, dn->nid, dn->inode->i_ino, ofs, true);
set_cold_node(page, S_ISDIR(dn->inode->i_mode));
if (!PageUptodate(page))
SetPageUptodate(page);
if (set_page_dirty(page))
dn->node_changed = true;
if (f2fs_has_xattr_block(ofs))
f2fs_i_xnid_write(dn->inode, dn->nid);
if (ofs == 0)
inc_valid_inode_count(sbi);
return page;
fail:
clear_node_page_dirty(page);
f2fs_put_page(page, 1);
return ERR_PTR(err);
}
/*
* Caller should do after getting the following values.
* 0: f2fs_put_page(page, 0)
* LOCKED_PAGE or error: f2fs_put_page(page, 1)
*/
static int read_node_page(struct page *page, int op_flags)
{
struct f2fs_sb_info *sbi = F2FS_P_SB(page);
struct node_info ni;
struct f2fs_io_info fio = {
.sbi = sbi,
.type = NODE,
.op = REQ_OP_READ,
.op_flags = op_flags,
.page = page,
.encrypted_page = NULL,
};
int err;
if (PageUptodate(page)) {
if (!f2fs_inode_chksum_verify(sbi, page)) {
ClearPageUptodate(page);
return -EFSBADCRC;
}
return LOCKED_PAGE;
}
err = f2fs_get_node_info(sbi, page->index, &ni);
if (err)
return err;
if (unlikely(ni.blk_addr == NULL_ADDR) ||
is_sbi_flag_set(sbi, SBI_IS_SHUTDOWN)) {
ClearPageUptodate(page);
return -ENOENT;
}
fio.new_blkaddr = fio.old_blkaddr = ni.blk_addr;
return f2fs_submit_page_bio(&fio);
}
/*
* Readahead a node 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
void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid)
{
struct page *apage;
int err;
if (!nid)
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
if (f2fs_check_nid_range(sbi, nid))
f2fs: give message and set need_fsck given broken node id 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=d154ec99402c6f628887 C reproducer: https://syzkaller.appspot.com/x/repro.c?id=5414336294027264 syzkaller reproducer: https://syzkaller.appspot.com/x/repro.syz?id=5471683234234368 Raw console output: https://syzkaller.appspot.com/x/log.txt?id=5436660795834368 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+d154ec99402c6f628887@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 ------------[ cut here ]------------ kernel BUG at fs/f2fs/node.c:1185! invalid opcode: 0000 [#1] SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) Modules linked in: CPU: 1 PID: 4549 Comm: syzkaller704305 Not tainted 4.17.0-rc1+ #10 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:__get_node_page+0xb68/0x16e0 fs/f2fs/node.c:1185 RSP: 0018:ffff8801d960e820 EFLAGS: 00010293 RAX: ffff8801d88205c0 RBX: 0000000000000003 RCX: ffffffff82f6cc06 RDX: 0000000000000000 RSI: ffffffff82f6d5e8 RDI: 0000000000000004 RBP: ffff8801d960ec30 R08: ffff8801d88205c0 R09: ffffed003b5e46c2 R10: 0000000000000003 R11: 0000000000000003 R12: ffff8801a86e00c0 R13: 0000000000000001 R14: ffff8801a86e0530 R15: ffff8801d9745240 FS: 000000000072c880(0000) GS:ffff8801daf00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f3d403209b8 CR3: 00000001d8f3f000 CR4: 00000000001406e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: get_node_page fs/f2fs/node.c:1237 [inline] truncate_xattr_node+0x152/0x2e0 fs/f2fs/node.c:1014 remove_inode_page+0x200/0xaf0 fs/f2fs/node.c:1039 f2fs_evict_inode+0xe86/0x1710 fs/f2fs/inode.c:547 evict+0x4a6/0x960 fs/inode.c:557 iput_final fs/inode.c:1519 [inline] iput+0x62d/0xa80 fs/inode.c:1545 f2fs_fill_super+0x5f4e/0x7bf0 fs/f2fs/super.c:2849 mount_bdev+0x30c/0x3e0 fs/super.c:1164 f2fs_mount+0x34/0x40 fs/f2fs/super.c:3020 mount_fs+0xae/0x328 fs/super.c:1267 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:2518 [inline] do_mount+0x564/0x3070 fs/namespace.c:2848 ksys_mount+0x12d/0x140 fs/namespace.c:3064 __do_sys_mount fs/namespace.c:3078 [inline] __se_sys_mount fs/namespace.c:3075 [inline] __x64_sys_mount+0xbe/0x150 fs/namespace.c:3075 do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x443dea RSP: 002b:00007ffcc7882368 EFLAGS: 00000297 ORIG_RAX: 00000000000000a5 RAX: ffffffffffffffda RBX: 0000000020000c00 RCX: 0000000000443dea RDX: 0000000020000000 RSI: 0000000020000100 RDI: 00007ffcc7882370 RBP: 0000000000000003 R08: 0000000020016a00 R09: 000000000000000a R10: 0000000000000000 R11: 0000000000000297 R12: 0000000000000004 R13: 0000000000402ce0 R14: 0000000000000000 R15: 0000000000000000 RIP: __get_node_page+0xb68/0x16e0 fs/f2fs/node.c:1185 RSP: ffff8801d960e820 ---[ end trace 4edbeb71f002bb76 ]--- Reported-and-tested-by: syzbot+d154ec99402c6f628887@syzkaller.appspotmail.com Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-04-24 05:02:31 +00:00
return;
apage = xa_load(&NODE_MAPPING(sbi)->i_pages, nid);
if (apage)
return;
apage = f2fs_grab_cache_page(NODE_MAPPING(sbi), nid, false);
if (!apage)
return;
err = read_node_page(apage, REQ_RAHEAD);
f2fs_put_page(apage, err ? 1 : 0);
}
static struct page *__get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid,
struct page *parent, int start)
{
struct page *page;
int err;
if (!nid)
return ERR_PTR(-ENOENT);
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_check_nid_range(sbi, nid))
f2fs: give message and set need_fsck given broken node id 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=d154ec99402c6f628887 C reproducer: https://syzkaller.appspot.com/x/repro.c?id=5414336294027264 syzkaller reproducer: https://syzkaller.appspot.com/x/repro.syz?id=5471683234234368 Raw console output: https://syzkaller.appspot.com/x/log.txt?id=5436660795834368 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+d154ec99402c6f628887@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 ------------[ cut here ]------------ kernel BUG at fs/f2fs/node.c:1185! invalid opcode: 0000 [#1] SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) Modules linked in: CPU: 1 PID: 4549 Comm: syzkaller704305 Not tainted 4.17.0-rc1+ #10 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:__get_node_page+0xb68/0x16e0 fs/f2fs/node.c:1185 RSP: 0018:ffff8801d960e820 EFLAGS: 00010293 RAX: ffff8801d88205c0 RBX: 0000000000000003 RCX: ffffffff82f6cc06 RDX: 0000000000000000 RSI: ffffffff82f6d5e8 RDI: 0000000000000004 RBP: ffff8801d960ec30 R08: ffff8801d88205c0 R09: ffffed003b5e46c2 R10: 0000000000000003 R11: 0000000000000003 R12: ffff8801a86e00c0 R13: 0000000000000001 R14: ffff8801a86e0530 R15: ffff8801d9745240 FS: 000000000072c880(0000) GS:ffff8801daf00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f3d403209b8 CR3: 00000001d8f3f000 CR4: 00000000001406e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: get_node_page fs/f2fs/node.c:1237 [inline] truncate_xattr_node+0x152/0x2e0 fs/f2fs/node.c:1014 remove_inode_page+0x200/0xaf0 fs/f2fs/node.c:1039 f2fs_evict_inode+0xe86/0x1710 fs/f2fs/inode.c:547 evict+0x4a6/0x960 fs/inode.c:557 iput_final fs/inode.c:1519 [inline] iput+0x62d/0xa80 fs/inode.c:1545 f2fs_fill_super+0x5f4e/0x7bf0 fs/f2fs/super.c:2849 mount_bdev+0x30c/0x3e0 fs/super.c:1164 f2fs_mount+0x34/0x40 fs/f2fs/super.c:3020 mount_fs+0xae/0x328 fs/super.c:1267 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:2518 [inline] do_mount+0x564/0x3070 fs/namespace.c:2848 ksys_mount+0x12d/0x140 fs/namespace.c:3064 __do_sys_mount fs/namespace.c:3078 [inline] __se_sys_mount fs/namespace.c:3075 [inline] __x64_sys_mount+0xbe/0x150 fs/namespace.c:3075 do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x443dea RSP: 002b:00007ffcc7882368 EFLAGS: 00000297 ORIG_RAX: 00000000000000a5 RAX: ffffffffffffffda RBX: 0000000020000c00 RCX: 0000000000443dea RDX: 0000000020000000 RSI: 0000000020000100 RDI: 00007ffcc7882370 RBP: 0000000000000003 R08: 0000000020016a00 R09: 000000000000000a R10: 0000000000000000 R11: 0000000000000297 R12: 0000000000000004 R13: 0000000000402ce0 R14: 0000000000000000 R15: 0000000000000000 RIP: __get_node_page+0xb68/0x16e0 fs/f2fs/node.c:1185 RSP: ffff8801d960e820 ---[ end trace 4edbeb71f002bb76 ]--- Reported-and-tested-by: syzbot+d154ec99402c6f628887@syzkaller.appspotmail.com Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-04-24 05:02:31 +00:00
return ERR_PTR(-EINVAL);
repeat:
page = f2fs_grab_cache_page(NODE_MAPPING(sbi), nid, false);
if (!page)
return ERR_PTR(-ENOMEM);
err = read_node_page(page, 0);
if (err < 0) {
f2fs_put_page(page, 1);
return ERR_PTR(err);
} else if (err == LOCKED_PAGE) {
err = 0;
goto page_hit;
}
if (parent)
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_node_pages(parent, start + 1, MAX_RA_NODE);
f2fs: give a chance to merge IOs by IO scheduler Previously, background GC submits many 4KB read requests to load victim blocks and/or its (i)node blocks. ... f2fs_gc : f2fs_readpage: ino = 1, page_index = 0xb61, blkaddr = 0x3b964ed f2fs_gc : block_rq_complete: 8,16 R () 499854968 + 8 [0] f2fs_gc : f2fs_readpage: ino = 1, page_index = 0xb6f, blkaddr = 0x3b964ee f2fs_gc : block_rq_complete: 8,16 R () 499854976 + 8 [0] f2fs_gc : f2fs_readpage: ino = 1, page_index = 0xb79, blkaddr = 0x3b964ef f2fs_gc : block_rq_complete: 8,16 R () 499854984 + 8 [0] ... However, by the fact that many IOs are sequential, we can give a chance to merge the IOs by IO scheduler. In order to do that, let's use blk_plug. ... f2fs_gc : f2fs_iget: ino = 143 f2fs_gc : f2fs_readpage: ino = 143, page_index = 0x1c6, blkaddr = 0x2e6ee f2fs_gc : f2fs_iget: ino = 143 f2fs_gc : f2fs_readpage: ino = 143, page_index = 0x1c7, blkaddr = 0x2e6ef <idle> : block_rq_complete: 8,16 R () 1519616 + 8 [0] <idle> : block_rq_complete: 8,16 R () 1519848 + 8 [0] <idle> : block_rq_complete: 8,16 R () 1520432 + 96 [0] <idle> : block_rq_complete: 8,16 R () 1520536 + 104 [0] <idle> : block_rq_complete: 8,16 R () 1521008 + 112 [0] <idle> : block_rq_complete: 8,16 R () 1521440 + 152 [0] <idle> : block_rq_complete: 8,16 R () 1521688 + 144 [0] <idle> : block_rq_complete: 8,16 R () 1522128 + 192 [0] <idle> : block_rq_complete: 8,16 R () 1523256 + 328 [0] ... Note that this issue should be addressed in checkpoint, and some readahead flows too. Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-04-24 04:19:56 +00:00
lock_page(page);
if (unlikely(page->mapping != NODE_MAPPING(sbi))) {
f2fs_put_page(page, 1);
goto repeat;
}
if (unlikely(!PageUptodate(page))) {
err = -EIO;
goto out_err;
}
if (!f2fs_inode_chksum_verify(sbi, page)) {
err = -EFSBADCRC;
goto out_err;
}
page_hit:
if(unlikely(nid != nid_of_node(page))) {
f2fs_warn(sbi, "inconsistent node block, nid:%lu, node_footer[nid:%u,ino:%u,ofs:%u,cpver:%llu,blkaddr:%u]",
nid, nid_of_node(page), ino_of_node(page),
ofs_of_node(page), cpver_of_node(page),
next_blkaddr_of_node(page));
err = -EINVAL;
out_err:
ClearPageUptodate(page);
f2fs_put_page(page, 1);
return ERR_PTR(err);
}
return 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_node_page(struct f2fs_sb_info *sbi, pgoff_t nid)
{
return __get_node_page(sbi, nid, 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
struct page *f2fs_get_node_page_ra(struct page *parent, int start)
{
struct f2fs_sb_info *sbi = F2FS_P_SB(parent);
nid_t nid = get_nid(parent, start, false);
return __get_node_page(sbi, nid, parent, start);
}
static void flush_inline_data(struct f2fs_sb_info *sbi, nid_t ino)
{
struct inode *inode;
struct page *page;
f2fs: fix to update dirty page count correctly Once we failed to merge inline data into inode page during flushing inline inode, we will skip invoking inode_dec_dirty_pages, which makes dirty page count incorrect, result in panic in ->evict_inode, Fix it. ------------[ cut here ]------------ kernel BUG at /home/yuchao/git/devf2fs/inode.c:336! invalid opcode: 0000 [#1] PREEMPT SMP CPU: 3 PID: 10004 Comm: umount Tainted: G O 4.6.0-rc5+ #17 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 task: f0c33000 ti: c5212000 task.ti: c5212000 EIP: 0060:[<f89aacb5>] EFLAGS: 00010202 CPU: 3 EIP is at f2fs_evict_inode+0x85/0x490 [f2fs] EAX: 00000001 EBX: c4529ea0 ECX: 00000001 EDX: 00000000 ESI: c0131000 EDI: f89dd0a0 EBP: c5213e9c ESP: c5213e78 DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 CR0: 80050033 CR2: b75878c0 CR3: 1a36a700 CR4: 000406f0 Stack: c4529ea0 c4529ef4 c5213e8c c176d45c c4529ef4 00000000 c4529ea0 c4529fac f89dd0a0 c5213eb0 c1204a68 c5213ed8 c452a2b4 c6680930 c5213ec0 c1204b64 c6680d44 c6680620 c5213eec c120588d ee84b000 ee84b5c0 c5214000 ee84b5e0 Call Trace: [<c176d45c>] ? _raw_spin_unlock+0x2c/0x50 [<c1204a68>] evict+0xa8/0x170 [<c1204b64>] dispose_list+0x34/0x50 [<c120588d>] evict_inodes+0x10d/0x130 [<c11ea941>] generic_shutdown_super+0x41/0xe0 [<c1185190>] ? unregister_shrinker+0x40/0x50 [<c1185190>] ? unregister_shrinker+0x40/0x50 [<c11eac52>] kill_block_super+0x22/0x70 [<f89af23e>] kill_f2fs_super+0x1e/0x20 [f2fs] [<c11eae1d>] deactivate_locked_super+0x3d/0x70 [<c11eb383>] deactivate_super+0x43/0x60 [<c1208ec9>] cleanup_mnt+0x39/0x80 [<c1208f50>] __cleanup_mnt+0x10/0x20 [<c107d091>] task_work_run+0x71/0x90 [<c105725a>] exit_to_usermode_loop+0x72/0x9e [<c1001c7c>] do_fast_syscall_32+0x19c/0x1c0 [<c176dd48>] sysenter_past_esp+0x45/0x74 EIP: [<f89aacb5>] f2fs_evict_inode+0x85/0x490 [f2fs] SS:ESP 0068:c5213e78 ---[ end trace d30536330b7fdc58 ]--- Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-05-20 16:11:09 +00:00
int ret;
/* should flush inline_data before evict_inode */
inode = ilookup(sbi->sb, ino);
if (!inode)
return;
page = f2fs_pagecache_get_page(inode->i_mapping, 0,
FGP_LOCK|FGP_NOWAIT, 0);
if (!page)
goto iput_out;
if (!PageUptodate(page))
goto page_out;
if (!PageDirty(page))
goto page_out;
if (!clear_page_dirty_for_io(page))
goto page_out;
f2fs: fix to update dirty page count correctly Once we failed to merge inline data into inode page during flushing inline inode, we will skip invoking inode_dec_dirty_pages, which makes dirty page count incorrect, result in panic in ->evict_inode, Fix it. ------------[ cut here ]------------ kernel BUG at /home/yuchao/git/devf2fs/inode.c:336! invalid opcode: 0000 [#1] PREEMPT SMP CPU: 3 PID: 10004 Comm: umount Tainted: G O 4.6.0-rc5+ #17 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 task: f0c33000 ti: c5212000 task.ti: c5212000 EIP: 0060:[<f89aacb5>] EFLAGS: 00010202 CPU: 3 EIP is at f2fs_evict_inode+0x85/0x490 [f2fs] EAX: 00000001 EBX: c4529ea0 ECX: 00000001 EDX: 00000000 ESI: c0131000 EDI: f89dd0a0 EBP: c5213e9c ESP: c5213e78 DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 CR0: 80050033 CR2: b75878c0 CR3: 1a36a700 CR4: 000406f0 Stack: c4529ea0 c4529ef4 c5213e8c c176d45c c4529ef4 00000000 c4529ea0 c4529fac f89dd0a0 c5213eb0 c1204a68 c5213ed8 c452a2b4 c6680930 c5213ec0 c1204b64 c6680d44 c6680620 c5213eec c120588d ee84b000 ee84b5c0 c5214000 ee84b5e0 Call Trace: [<c176d45c>] ? _raw_spin_unlock+0x2c/0x50 [<c1204a68>] evict+0xa8/0x170 [<c1204b64>] dispose_list+0x34/0x50 [<c120588d>] evict_inodes+0x10d/0x130 [<c11ea941>] generic_shutdown_super+0x41/0xe0 [<c1185190>] ? unregister_shrinker+0x40/0x50 [<c1185190>] ? unregister_shrinker+0x40/0x50 [<c11eac52>] kill_block_super+0x22/0x70 [<f89af23e>] kill_f2fs_super+0x1e/0x20 [f2fs] [<c11eae1d>] deactivate_locked_super+0x3d/0x70 [<c11eb383>] deactivate_super+0x43/0x60 [<c1208ec9>] cleanup_mnt+0x39/0x80 [<c1208f50>] __cleanup_mnt+0x10/0x20 [<c107d091>] task_work_run+0x71/0x90 [<c105725a>] exit_to_usermode_loop+0x72/0x9e [<c1001c7c>] do_fast_syscall_32+0x19c/0x1c0 [<c176dd48>] sysenter_past_esp+0x45/0x74 EIP: [<f89aacb5>] f2fs_evict_inode+0x85/0x490 [f2fs] SS:ESP 0068:c5213e78 ---[ end trace d30536330b7fdc58 ]--- Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-05-20 16:11:09 +00:00
ret = f2fs_write_inline_data(inode, 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);
f2fs: fix to update dirty page count correctly Once we failed to merge inline data into inode page during flushing inline inode, we will skip invoking inode_dec_dirty_pages, which makes dirty page count incorrect, result in panic in ->evict_inode, Fix it. ------------[ cut here ]------------ kernel BUG at /home/yuchao/git/devf2fs/inode.c:336! invalid opcode: 0000 [#1] PREEMPT SMP CPU: 3 PID: 10004 Comm: umount Tainted: G O 4.6.0-rc5+ #17 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 task: f0c33000 ti: c5212000 task.ti: c5212000 EIP: 0060:[<f89aacb5>] EFLAGS: 00010202 CPU: 3 EIP is at f2fs_evict_inode+0x85/0x490 [f2fs] EAX: 00000001 EBX: c4529ea0 ECX: 00000001 EDX: 00000000 ESI: c0131000 EDI: f89dd0a0 EBP: c5213e9c ESP: c5213e78 DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 CR0: 80050033 CR2: b75878c0 CR3: 1a36a700 CR4: 000406f0 Stack: c4529ea0 c4529ef4 c5213e8c c176d45c c4529ef4 00000000 c4529ea0 c4529fac f89dd0a0 c5213eb0 c1204a68 c5213ed8 c452a2b4 c6680930 c5213ec0 c1204b64 c6680d44 c6680620 c5213eec c120588d ee84b000 ee84b5c0 c5214000 ee84b5e0 Call Trace: [<c176d45c>] ? _raw_spin_unlock+0x2c/0x50 [<c1204a68>] evict+0xa8/0x170 [<c1204b64>] dispose_list+0x34/0x50 [<c120588d>] evict_inodes+0x10d/0x130 [<c11ea941>] generic_shutdown_super+0x41/0xe0 [<c1185190>] ? unregister_shrinker+0x40/0x50 [<c1185190>] ? unregister_shrinker+0x40/0x50 [<c11eac52>] kill_block_super+0x22/0x70 [<f89af23e>] kill_f2fs_super+0x1e/0x20 [f2fs] [<c11eae1d>] deactivate_locked_super+0x3d/0x70 [<c11eb383>] deactivate_super+0x43/0x60 [<c1208ec9>] cleanup_mnt+0x39/0x80 [<c1208f50>] __cleanup_mnt+0x10/0x20 [<c107d091>] task_work_run+0x71/0x90 [<c105725a>] exit_to_usermode_loop+0x72/0x9e [<c1001c7c>] do_fast_syscall_32+0x19c/0x1c0 [<c176dd48>] sysenter_past_esp+0x45/0x74 EIP: [<f89aacb5>] f2fs_evict_inode+0x85/0x490 [f2fs] SS:ESP 0068:c5213e78 ---[ end trace d30536330b7fdc58 ]--- Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-05-20 16:11:09 +00:00
if (ret)
set_page_dirty(page);
page_out:
f2fs_put_page(page, 1);
iput_out:
iput(inode);
}
static struct page *last_fsync_dnode(struct f2fs_sb_info *sbi, nid_t ino)
{
pgoff_t index;
struct pagevec pvec;
struct page *last_page = NULL;
int nr_pages;
pagevec_init(&pvec);
index = 0;
while ((nr_pages = pagevec_lookup_tag(&pvec, NODE_MAPPING(sbi), &index,
PAGECACHE_TAG_DIRTY))) {
int i;
for (i = 0; i < nr_pages; i++) {
struct page *page = pvec.pages[i];
if (unlikely(f2fs_cp_error(sbi))) {
f2fs_put_page(last_page, 0);
pagevec_release(&pvec);
return ERR_PTR(-EIO);
}
if (!IS_DNODE(page) || !is_cold_node(page))
continue;
if (ino_of_node(page) != ino)
continue;
lock_page(page);
if (unlikely(page->mapping != NODE_MAPPING(sbi))) {
continue_unlock:
unlock_page(page);
continue;
}
if (ino_of_node(page) != ino)
goto continue_unlock;
if (!PageDirty(page)) {
/* someone wrote it for us */
goto continue_unlock;
}
if (last_page)
f2fs_put_page(last_page, 0);
get_page(page);
last_page = page;
unlock_page(page);
}
pagevec_release(&pvec);
cond_resched();
}
return last_page;
}
static int __write_node_page(struct page *page, bool atomic, bool *submitted,
struct writeback_control *wbc, bool do_balance,
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
enum iostat_type io_type, unsigned int *seq_id)
{
struct f2fs_sb_info *sbi = F2FS_P_SB(page);
nid_t nid;
struct node_info ni;
struct f2fs_io_info fio = {
.sbi = sbi,
.ino = ino_of_node(page),
.type = NODE,
.op = REQ_OP_WRITE,
.op_flags = wbc_to_write_flags(wbc),
.page = page,
.encrypted_page = NULL,
.submitted = false,
.io_type = io_type,
.io_wbc = wbc,
};
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
unsigned int seq;
trace_f2fs_writepage(page, NODE);
if (unlikely(f2fs_cp_error(sbi)))
goto redirty_out;
if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
goto redirty_out;
if (!is_sbi_flag_set(sbi, SBI_CP_DISABLED) &&
wbc->sync_mode == WB_SYNC_NONE &&
IS_DNODE(page) && is_cold_node(page))
goto redirty_out;
/* get old block addr of this node page */
nid = nid_of_node(page);
f2fs_bug_on(sbi, page->index != nid);
if (f2fs_get_node_info(sbi, nid, &ni))
goto redirty_out;
if (wbc->for_reclaim) {
if (!down_read_trylock(&sbi->node_write))
goto redirty_out;
} else {
down_read(&sbi->node_write);
}
/* This page is already truncated */
if (unlikely(ni.blk_addr == NULL_ADDR)) {
ClearPageUptodate(page);
dec_page_count(sbi, F2FS_DIRTY_NODES);
up_read(&sbi->node_write);
unlock_page(page);
return 0;
}
f2fs: fix to do sanity check with block address in main area This patch add to do sanity check with below field: - cp_pack_total_block_count - blkaddr of data/node - extent info - Overview BUG() in verify_block_addr() when writing to a corrupted f2fs image - Reproduce (4.18 upstream kernel) - POC (poc.c) static void activity(char *mpoint) { char *foo_bar_baz; int err; static int buf[8192]; memset(buf, 0, sizeof(buf)); err = asprintf(&foo_bar_baz, "%s/foo/bar/baz", mpoint); int fd = open(foo_bar_baz, O_RDWR | O_TRUNC, 0777); if (fd >= 0) { write(fd, (char *)buf, sizeof(buf)); fdatasync(fd); close(fd); } } int main(int argc, char *argv[]) { activity(argv[1]); return 0; } - Kernel message [ 689.349473] F2FS-fs (loop0): Mounted with checkpoint version = 3 [ 699.728662] WARNING: CPU: 0 PID: 1309 at fs/f2fs/segment.c:2860 f2fs_inplace_write_data+0x232/0x240 [ 699.728670] Modules linked in: snd_hda_codec_generic snd_hda_intel snd_hda_codec snd_hwdep snd_hda_core snd_pcm snd_timer snd mac_hid i2c_piix4 soundcore ib_iser rdma_cm iw_cm ib_cm ib_core iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx raid1 raid0 multipath linear 8139too crct10dif_pclmul crc32_pclmul qxl drm_kms_helper syscopyarea aesni_intel sysfillrect sysimgblt fb_sys_fops ttm drm aes_x86_64 crypto_simd cryptd 8139cp glue_helper mii pata_acpi floppy [ 699.729056] CPU: 0 PID: 1309 Comm: a.out Not tainted 4.18.0-rc1+ #4 [ 699.729064] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014 [ 699.729074] RIP: 0010:f2fs_inplace_write_data+0x232/0x240 [ 699.729076] Code: ff e9 cf fe ff ff 49 8d 7d 10 e8 39 45 ad ff 4d 8b 7d 10 be 04 00 00 00 49 8d 7f 48 e8 07 49 ad ff 45 8b 7f 48 e9 fb fe ff ff <0f> 0b f0 41 80 4d 48 04 e9 65 fe ff ff 90 66 66 66 66 90 55 48 8d [ 699.729130] RSP: 0018:ffff8801f43af568 EFLAGS: 00010202 [ 699.729139] RAX: 000000000000003f RBX: ffff8801f43af7b8 RCX: ffffffffb88c9113 [ 699.729142] RDX: 0000000000000003 RSI: dffffc0000000000 RDI: ffff8802024e5540 [ 699.729144] RBP: ffff8801f43af590 R08: 0000000000000009 R09: ffffffffffffffe8 [ 699.729147] R10: 0000000000000001 R11: ffffed0039b0596a R12: ffff8802024e5540 [ 699.729149] R13: ffff8801f0335500 R14: ffff8801e3e7a700 R15: ffff8801e1ee4450 [ 699.729154] FS: 00007f9bf97f5700(0000) GS:ffff8801f6e00000(0000) knlGS:0000000000000000 [ 699.729156] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 699.729159] CR2: 00007f9bf925d170 CR3: 00000001f0c34000 CR4: 00000000000006f0 [ 699.729171] Call Trace: [ 699.729192] f2fs_do_write_data_page+0x2e2/0xe00 [ 699.729203] ? f2fs_should_update_outplace+0xd0/0xd0 [ 699.729238] ? memcg_drain_all_list_lrus+0x280/0x280 [ 699.729269] ? __radix_tree_replace+0xa3/0x120 [ 699.729276] __write_data_page+0x5c7/0xe30 [ 699.729291] ? kasan_check_read+0x11/0x20 [ 699.729310] ? page_mapped+0x8a/0x110 [ 699.729321] ? page_mkclean+0xe9/0x160 [ 699.729327] ? f2fs_do_write_data_page+0xe00/0xe00 [ 699.729331] ? invalid_page_referenced_vma+0x130/0x130 [ 699.729345] ? clear_page_dirty_for_io+0x332/0x450 [ 699.729351] f2fs_write_cache_pages+0x4ca/0x860 [ 699.729358] ? __write_data_page+0xe30/0xe30 [ 699.729374] ? percpu_counter_add_batch+0x22/0xa0 [ 699.729380] ? kasan_check_write+0x14/0x20 [ 699.729391] ? _raw_spin_lock+0x17/0x40 [ 699.729403] ? f2fs_mark_inode_dirty_sync.part.18+0x16/0x30 [ 699.729413] ? iov_iter_advance+0x113/0x640 [ 699.729418] ? f2fs_write_end+0x133/0x2e0 [ 699.729423] ? balance_dirty_pages_ratelimited+0x239/0x640 [ 699.729428] f2fs_write_data_pages+0x329/0x520 [ 699.729433] ? generic_perform_write+0x250/0x320 [ 699.729438] ? f2fs_write_cache_pages+0x860/0x860 [ 699.729454] ? current_time+0x110/0x110 [ 699.729459] ? f2fs_preallocate_blocks+0x1ef/0x370 [ 699.729464] do_writepages+0x37/0xb0 [ 699.729468] ? f2fs_write_cache_pages+0x860/0x860 [ 699.729472] ? do_writepages+0x37/0xb0 [ 699.729478] __filemap_fdatawrite_range+0x19a/0x1f0 [ 699.729483] ? delete_from_page_cache_batch+0x4e0/0x4e0 [ 699.729496] ? __vfs_write+0x2b2/0x410 [ 699.729501] file_write_and_wait_range+0x66/0xb0 [ 699.729506] f2fs_do_sync_file+0x1f9/0xd90 [ 699.729511] ? truncate_partial_data_page+0x290/0x290 [ 699.729521] ? __sb_end_write+0x30/0x50 [ 699.729526] ? vfs_write+0x20f/0x260 [ 699.729530] f2fs_sync_file+0x9a/0xb0 [ 699.729534] ? f2fs_do_sync_file+0xd90/0xd90 [ 699.729548] vfs_fsync_range+0x68/0x100 [ 699.729554] ? __fget_light+0xc9/0xe0 [ 699.729558] do_fsync+0x3d/0x70 [ 699.729562] __x64_sys_fdatasync+0x24/0x30 [ 699.729585] do_syscall_64+0x78/0x170 [ 699.729595] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 699.729613] RIP: 0033:0x7f9bf930d800 [ 699.729615] Code: 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 83 3d 49 bf 2c 00 00 75 10 b8 4b 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 31 c3 48 83 ec 08 e8 be 78 01 00 48 89 04 24 [ 699.729668] RSP: 002b:00007ffee3606c68 EFLAGS: 00000246 ORIG_RAX: 000000000000004b [ 699.729673] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f9bf930d800 [ 699.729675] RDX: 0000000000008000 RSI: 00000000006010a0 RDI: 0000000000000003 [ 699.729678] RBP: 00007ffee3606ca0 R08: 0000000001503010 R09: 0000000000000000 [ 699.729680] R10: 00000000000002e8 R11: 0000000000000246 R12: 0000000000400610 [ 699.729683] R13: 00007ffee3606da0 R14: 0000000000000000 R15: 0000000000000000 [ 699.729687] ---[ end trace 4ce02f25ff7d3df5 ]--- [ 699.729782] ------------[ cut here ]------------ [ 699.729785] kernel BUG at fs/f2fs/segment.h:654! [ 699.731055] invalid opcode: 0000 [#1] SMP KASAN PTI [ 699.732104] CPU: 0 PID: 1309 Comm: a.out Tainted: G W 4.18.0-rc1+ #4 [ 699.733684] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014 [ 699.735611] RIP: 0010:f2fs_submit_page_bio+0x29b/0x730 [ 699.736649] Code: 54 49 8d bd 18 04 00 00 e8 b2 59 af ff 41 8b 8d 18 04 00 00 8b 45 b8 41 d3 e6 44 01 f0 4c 8d 73 14 41 39 c7 0f 82 37 fe ff ff <0f> 0b 65 8b 05 2c 04 77 47 89 c0 48 0f a3 05 52 c1 d5 01 0f 92 c0 [ 699.740524] RSP: 0018:ffff8801f43af508 EFLAGS: 00010283 [ 699.741573] RAX: 0000000000000000 RBX: ffff8801f43af7b8 RCX: ffffffffb88a7cef [ 699.743006] RDX: 0000000000000007 RSI: dffffc0000000000 RDI: ffff8801e3e7a64c [ 699.744426] RBP: ffff8801f43af558 R08: ffffed003e066b55 R09: ffffed003e066b55 [ 699.745833] R10: 0000000000000001 R11: ffffed003e066b54 R12: ffffea0007876940 [ 699.747256] R13: ffff8801f0335500 R14: ffff8801e3e7a600 R15: 0000000000000001 [ 699.748683] FS: 00007f9bf97f5700(0000) GS:ffff8801f6e00000(0000) knlGS:0000000000000000 [ 699.750293] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 699.751462] CR2: 00007f9bf925d170 CR3: 00000001f0c34000 CR4: 00000000000006f0 [ 699.752874] Call Trace: [ 699.753386] ? f2fs_inplace_write_data+0x93/0x240 [ 699.754341] f2fs_inplace_write_data+0xd2/0x240 [ 699.755271] f2fs_do_write_data_page+0x2e2/0xe00 [ 699.756214] ? f2fs_should_update_outplace+0xd0/0xd0 [ 699.757215] ? memcg_drain_all_list_lrus+0x280/0x280 [ 699.758209] ? __radix_tree_replace+0xa3/0x120 [ 699.759164] __write_data_page+0x5c7/0xe30 [ 699.760002] ? kasan_check_read+0x11/0x20 [ 699.760823] ? page_mapped+0x8a/0x110 [ 699.761573] ? page_mkclean+0xe9/0x160 [ 699.762345] ? f2fs_do_write_data_page+0xe00/0xe00 [ 699.763332] ? invalid_page_referenced_vma+0x130/0x130 [ 699.764374] ? clear_page_dirty_for_io+0x332/0x450 [ 699.765347] f2fs_write_cache_pages+0x4ca/0x860 [ 699.766276] ? __write_data_page+0xe30/0xe30 [ 699.767161] ? percpu_counter_add_batch+0x22/0xa0 [ 699.768112] ? kasan_check_write+0x14/0x20 [ 699.768951] ? _raw_spin_lock+0x17/0x40 [ 699.769739] ? f2fs_mark_inode_dirty_sync.part.18+0x16/0x30 [ 699.770885] ? iov_iter_advance+0x113/0x640 [ 699.771743] ? f2fs_write_end+0x133/0x2e0 [ 699.772569] ? balance_dirty_pages_ratelimited+0x239/0x640 [ 699.773680] f2fs_write_data_pages+0x329/0x520 [ 699.774603] ? generic_perform_write+0x250/0x320 [ 699.775544] ? f2fs_write_cache_pages+0x860/0x860 [ 699.776510] ? current_time+0x110/0x110 [ 699.777299] ? f2fs_preallocate_blocks+0x1ef/0x370 [ 699.778279] do_writepages+0x37/0xb0 [ 699.779026] ? f2fs_write_cache_pages+0x860/0x860 [ 699.779978] ? do_writepages+0x37/0xb0 [ 699.780755] __filemap_fdatawrite_range+0x19a/0x1f0 [ 699.781746] ? delete_from_page_cache_batch+0x4e0/0x4e0 [ 699.782820] ? __vfs_write+0x2b2/0x410 [ 699.783597] file_write_and_wait_range+0x66/0xb0 [ 699.784540] f2fs_do_sync_file+0x1f9/0xd90 [ 699.785381] ? truncate_partial_data_page+0x290/0x290 [ 699.786415] ? __sb_end_write+0x30/0x50 [ 699.787204] ? vfs_write+0x20f/0x260 [ 699.787941] f2fs_sync_file+0x9a/0xb0 [ 699.788694] ? f2fs_do_sync_file+0xd90/0xd90 [ 699.789572] vfs_fsync_range+0x68/0x100 [ 699.790360] ? __fget_light+0xc9/0xe0 [ 699.791128] do_fsync+0x3d/0x70 [ 699.791779] __x64_sys_fdatasync+0x24/0x30 [ 699.792614] do_syscall_64+0x78/0x170 [ 699.793371] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 699.794406] RIP: 0033:0x7f9bf930d800 [ 699.795134] Code: 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 83 3d 49 bf 2c 00 00 75 10 b8 4b 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 31 c3 48 83 ec 08 e8 be 78 01 00 48 89 04 24 [ 699.798960] RSP: 002b:00007ffee3606c68 EFLAGS: 00000246 ORIG_RAX: 000000000000004b [ 699.800483] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f9bf930d800 [ 699.801923] RDX: 0000000000008000 RSI: 00000000006010a0 RDI: 0000000000000003 [ 699.803373] RBP: 00007ffee3606ca0 R08: 0000000001503010 R09: 0000000000000000 [ 699.804798] R10: 00000000000002e8 R11: 0000000000000246 R12: 0000000000400610 [ 699.806233] R13: 00007ffee3606da0 R14: 0000000000000000 R15: 0000000000000000 [ 699.807667] Modules linked in: snd_hda_codec_generic snd_hda_intel snd_hda_codec snd_hwdep snd_hda_core snd_pcm snd_timer snd mac_hid i2c_piix4 soundcore ib_iser rdma_cm iw_cm ib_cm ib_core iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx raid1 raid0 multipath linear 8139too crct10dif_pclmul crc32_pclmul qxl drm_kms_helper syscopyarea aesni_intel sysfillrect sysimgblt fb_sys_fops ttm drm aes_x86_64 crypto_simd cryptd 8139cp glue_helper mii pata_acpi floppy [ 699.817079] ---[ end trace 4ce02f25ff7d3df6 ]--- [ 699.818068] RIP: 0010:f2fs_submit_page_bio+0x29b/0x730 [ 699.819114] Code: 54 49 8d bd 18 04 00 00 e8 b2 59 af ff 41 8b 8d 18 04 00 00 8b 45 b8 41 d3 e6 44 01 f0 4c 8d 73 14 41 39 c7 0f 82 37 fe ff ff <0f> 0b 65 8b 05 2c 04 77 47 89 c0 48 0f a3 05 52 c1 d5 01 0f 92 c0 [ 699.822919] RSP: 0018:ffff8801f43af508 EFLAGS: 00010283 [ 699.823977] RAX: 0000000000000000 RBX: ffff8801f43af7b8 RCX: ffffffffb88a7cef [ 699.825436] RDX: 0000000000000007 RSI: dffffc0000000000 RDI: ffff8801e3e7a64c [ 699.826881] RBP: ffff8801f43af558 R08: ffffed003e066b55 R09: ffffed003e066b55 [ 699.828292] R10: 0000000000000001 R11: ffffed003e066b54 R12: ffffea0007876940 [ 699.829750] R13: ffff8801f0335500 R14: ffff8801e3e7a600 R15: 0000000000000001 [ 699.831192] FS: 00007f9bf97f5700(0000) GS:ffff8801f6e00000(0000) knlGS:0000000000000000 [ 699.832793] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 699.833981] CR2: 00007f9bf925d170 CR3: 00000001f0c34000 CR4: 00000000000006f0 [ 699.835556] ================================================================== [ 699.837029] BUG: KASAN: stack-out-of-bounds in update_stack_state+0x38c/0x3e0 [ 699.838462] Read of size 8 at addr ffff8801f43af970 by task a.out/1309 [ 699.840086] CPU: 0 PID: 1309 Comm: a.out Tainted: G D W 4.18.0-rc1+ #4 [ 699.841603] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014 [ 699.843475] Call Trace: [ 699.843982] dump_stack+0x7b/0xb5 [ 699.844661] print_address_description+0x70/0x290 [ 699.845607] kasan_report+0x291/0x390 [ 699.846351] ? update_stack_state+0x38c/0x3e0 [ 699.853831] __asan_load8+0x54/0x90 [ 699.854569] update_stack_state+0x38c/0x3e0 [ 699.855428] ? __read_once_size_nocheck.constprop.7+0x20/0x20 [ 699.856601] ? __save_stack_trace+0x5e/0x100 [ 699.857476] unwind_next_frame.part.5+0x18e/0x490 [ 699.858448] ? unwind_dump+0x290/0x290 [ 699.859217] ? clear_page_dirty_for_io+0x332/0x450 [ 699.860185] __unwind_start+0x106/0x190 [ 699.860974] __save_stack_trace+0x5e/0x100 [ 699.861808] ? __save_stack_trace+0x5e/0x100 [ 699.862691] ? unlink_anon_vmas+0xba/0x2c0 [ 699.863525] save_stack_trace+0x1f/0x30 [ 699.864312] save_stack+0x46/0xd0 [ 699.864993] ? __alloc_pages_slowpath+0x1420/0x1420 [ 699.865990] ? flush_tlb_mm_range+0x15e/0x220 [ 699.866889] ? kasan_check_write+0x14/0x20 [ 699.867724] ? __dec_node_state+0x92/0xb0 [ 699.868543] ? lock_page_memcg+0x85/0xf0 [ 699.869350] ? unlock_page_memcg+0x16/0x80 [ 699.870185] ? page_remove_rmap+0x198/0x520 [ 699.871048] ? mark_page_accessed+0x133/0x200 [ 699.871930] ? _cond_resched+0x1a/0x50 [ 699.872700] ? unmap_page_range+0xcd4/0xe50 [ 699.873551] ? rb_next+0x58/0x80 [ 699.874217] ? rb_next+0x58/0x80 [ 699.874895] __kasan_slab_free+0x13c/0x1a0 [ 699.875734] ? unlink_anon_vmas+0xba/0x2c0 [ 699.876563] kasan_slab_free+0xe/0x10 [ 699.877315] kmem_cache_free+0x89/0x1e0 [ 699.878095] unlink_anon_vmas+0xba/0x2c0 [ 699.878913] free_pgtables+0x101/0x1b0 [ 699.879677] exit_mmap+0x146/0x2a0 [ 699.880378] ? __ia32_sys_munmap+0x50/0x50 [ 699.881214] ? kasan_check_read+0x11/0x20 [ 699.882052] ? mm_update_next_owner+0x322/0x380 [ 699.882985] mmput+0x8b/0x1d0 [ 699.883602] do_exit+0x43a/0x1390 [ 699.884288] ? mm_update_next_owner+0x380/0x380 [ 699.885212] ? f2fs_sync_file+0x9a/0xb0 [ 699.885995] ? f2fs_do_sync_file+0xd90/0xd90 [ 699.886877] ? vfs_fsync_range+0x68/0x100 [ 699.887694] ? __fget_light+0xc9/0xe0 [ 699.888442] ? do_fsync+0x3d/0x70 [ 699.889118] ? __x64_sys_fdatasync+0x24/0x30 [ 699.889996] rewind_stack_do_exit+0x17/0x20 [ 699.890860] RIP: 0033:0x7f9bf930d800 [ 699.891585] Code: Bad RIP value. [ 699.892268] RSP: 002b:00007ffee3606c68 EFLAGS: 00000246 ORIG_RAX: 000000000000004b [ 699.893781] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f9bf930d800 [ 699.895220] RDX: 0000000000008000 RSI: 00000000006010a0 RDI: 0000000000000003 [ 699.896643] RBP: 00007ffee3606ca0 R08: 0000000001503010 R09: 0000000000000000 [ 699.898069] R10: 00000000000002e8 R11: 0000000000000246 R12: 0000000000400610 [ 699.899505] R13: 00007ffee3606da0 R14: 0000000000000000 R15: 0000000000000000 [ 699.901241] The buggy address belongs to the page: [ 699.902215] page:ffffea0007d0ebc0 count:0 mapcount:0 mapping:0000000000000000 index:0x0 [ 699.903811] flags: 0x2ffff0000000000() [ 699.904585] raw: 02ffff0000000000 0000000000000000 ffffffff07d00101 0000000000000000 [ 699.906125] raw: 0000000000000000 0000000000240000 00000000ffffffff 0000000000000000 [ 699.907673] page dumped because: kasan: bad access detected [ 699.909108] Memory state around the buggy address: [ 699.910077] ffff8801f43af800: 00 f1 f1 f1 f1 00 f4 f4 f4 f3 f3 f3 f3 00 00 00 [ 699.911528] ffff8801f43af880: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 [ 699.912953] >ffff8801f43af900: 00 00 00 00 00 00 00 00 f1 01 f4 f4 f4 f2 f2 f2 [ 699.914392] ^ [ 699.915758] ffff8801f43af980: f2 00 f4 f4 00 00 00 00 f2 00 00 00 00 00 00 00 [ 699.917193] ffff8801f43afa00: 00 00 00 00 00 00 00 00 00 f3 f3 f3 00 00 00 00 [ 699.918634] ================================================================== - Location https://elixir.bootlin.com/linux/v4.18-rc1/source/fs/f2fs/segment.h#L644 Reported-by Wen Xu <wen.xu@gatech.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-01 11:13:44 +00:00
if (__is_valid_data_blkaddr(ni.blk_addr) &&
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
!f2fs_is_valid_blkaddr(sbi, ni.blk_addr,
DATA_GENERIC_ENHANCE)) {
up_read(&sbi->node_write);
f2fs: fix to do sanity check with block address in main area This patch add to do sanity check with below field: - cp_pack_total_block_count - blkaddr of data/node - extent info - Overview BUG() in verify_block_addr() when writing to a corrupted f2fs image - Reproduce (4.18 upstream kernel) - POC (poc.c) static void activity(char *mpoint) { char *foo_bar_baz; int err; static int buf[8192]; memset(buf, 0, sizeof(buf)); err = asprintf(&foo_bar_baz, "%s/foo/bar/baz", mpoint); int fd = open(foo_bar_baz, O_RDWR | O_TRUNC, 0777); if (fd >= 0) { write(fd, (char *)buf, sizeof(buf)); fdatasync(fd); close(fd); } } int main(int argc, char *argv[]) { activity(argv[1]); return 0; } - Kernel message [ 689.349473] F2FS-fs (loop0): Mounted with checkpoint version = 3 [ 699.728662] WARNING: CPU: 0 PID: 1309 at fs/f2fs/segment.c:2860 f2fs_inplace_write_data+0x232/0x240 [ 699.728670] Modules linked in: snd_hda_codec_generic snd_hda_intel snd_hda_codec snd_hwdep snd_hda_core snd_pcm snd_timer snd mac_hid i2c_piix4 soundcore ib_iser rdma_cm iw_cm ib_cm ib_core iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx raid1 raid0 multipath linear 8139too crct10dif_pclmul crc32_pclmul qxl drm_kms_helper syscopyarea aesni_intel sysfillrect sysimgblt fb_sys_fops ttm drm aes_x86_64 crypto_simd cryptd 8139cp glue_helper mii pata_acpi floppy [ 699.729056] CPU: 0 PID: 1309 Comm: a.out Not tainted 4.18.0-rc1+ #4 [ 699.729064] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014 [ 699.729074] RIP: 0010:f2fs_inplace_write_data+0x232/0x240 [ 699.729076] Code: ff e9 cf fe ff ff 49 8d 7d 10 e8 39 45 ad ff 4d 8b 7d 10 be 04 00 00 00 49 8d 7f 48 e8 07 49 ad ff 45 8b 7f 48 e9 fb fe ff ff <0f> 0b f0 41 80 4d 48 04 e9 65 fe ff ff 90 66 66 66 66 90 55 48 8d [ 699.729130] RSP: 0018:ffff8801f43af568 EFLAGS: 00010202 [ 699.729139] RAX: 000000000000003f RBX: ffff8801f43af7b8 RCX: ffffffffb88c9113 [ 699.729142] RDX: 0000000000000003 RSI: dffffc0000000000 RDI: ffff8802024e5540 [ 699.729144] RBP: ffff8801f43af590 R08: 0000000000000009 R09: ffffffffffffffe8 [ 699.729147] R10: 0000000000000001 R11: ffffed0039b0596a R12: ffff8802024e5540 [ 699.729149] R13: ffff8801f0335500 R14: ffff8801e3e7a700 R15: ffff8801e1ee4450 [ 699.729154] FS: 00007f9bf97f5700(0000) GS:ffff8801f6e00000(0000) knlGS:0000000000000000 [ 699.729156] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 699.729159] CR2: 00007f9bf925d170 CR3: 00000001f0c34000 CR4: 00000000000006f0 [ 699.729171] Call Trace: [ 699.729192] f2fs_do_write_data_page+0x2e2/0xe00 [ 699.729203] ? f2fs_should_update_outplace+0xd0/0xd0 [ 699.729238] ? memcg_drain_all_list_lrus+0x280/0x280 [ 699.729269] ? __radix_tree_replace+0xa3/0x120 [ 699.729276] __write_data_page+0x5c7/0xe30 [ 699.729291] ? kasan_check_read+0x11/0x20 [ 699.729310] ? page_mapped+0x8a/0x110 [ 699.729321] ? page_mkclean+0xe9/0x160 [ 699.729327] ? f2fs_do_write_data_page+0xe00/0xe00 [ 699.729331] ? invalid_page_referenced_vma+0x130/0x130 [ 699.729345] ? clear_page_dirty_for_io+0x332/0x450 [ 699.729351] f2fs_write_cache_pages+0x4ca/0x860 [ 699.729358] ? __write_data_page+0xe30/0xe30 [ 699.729374] ? percpu_counter_add_batch+0x22/0xa0 [ 699.729380] ? kasan_check_write+0x14/0x20 [ 699.729391] ? _raw_spin_lock+0x17/0x40 [ 699.729403] ? f2fs_mark_inode_dirty_sync.part.18+0x16/0x30 [ 699.729413] ? iov_iter_advance+0x113/0x640 [ 699.729418] ? f2fs_write_end+0x133/0x2e0 [ 699.729423] ? balance_dirty_pages_ratelimited+0x239/0x640 [ 699.729428] f2fs_write_data_pages+0x329/0x520 [ 699.729433] ? generic_perform_write+0x250/0x320 [ 699.729438] ? f2fs_write_cache_pages+0x860/0x860 [ 699.729454] ? current_time+0x110/0x110 [ 699.729459] ? f2fs_preallocate_blocks+0x1ef/0x370 [ 699.729464] do_writepages+0x37/0xb0 [ 699.729468] ? f2fs_write_cache_pages+0x860/0x860 [ 699.729472] ? do_writepages+0x37/0xb0 [ 699.729478] __filemap_fdatawrite_range+0x19a/0x1f0 [ 699.729483] ? delete_from_page_cache_batch+0x4e0/0x4e0 [ 699.729496] ? __vfs_write+0x2b2/0x410 [ 699.729501] file_write_and_wait_range+0x66/0xb0 [ 699.729506] f2fs_do_sync_file+0x1f9/0xd90 [ 699.729511] ? truncate_partial_data_page+0x290/0x290 [ 699.729521] ? __sb_end_write+0x30/0x50 [ 699.729526] ? vfs_write+0x20f/0x260 [ 699.729530] f2fs_sync_file+0x9a/0xb0 [ 699.729534] ? f2fs_do_sync_file+0xd90/0xd90 [ 699.729548] vfs_fsync_range+0x68/0x100 [ 699.729554] ? __fget_light+0xc9/0xe0 [ 699.729558] do_fsync+0x3d/0x70 [ 699.729562] __x64_sys_fdatasync+0x24/0x30 [ 699.729585] do_syscall_64+0x78/0x170 [ 699.729595] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 699.729613] RIP: 0033:0x7f9bf930d800 [ 699.729615] Code: 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 83 3d 49 bf 2c 00 00 75 10 b8 4b 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 31 c3 48 83 ec 08 e8 be 78 01 00 48 89 04 24 [ 699.729668] RSP: 002b:00007ffee3606c68 EFLAGS: 00000246 ORIG_RAX: 000000000000004b [ 699.729673] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f9bf930d800 [ 699.729675] RDX: 0000000000008000 RSI: 00000000006010a0 RDI: 0000000000000003 [ 699.729678] RBP: 00007ffee3606ca0 R08: 0000000001503010 R09: 0000000000000000 [ 699.729680] R10: 00000000000002e8 R11: 0000000000000246 R12: 0000000000400610 [ 699.729683] R13: 00007ffee3606da0 R14: 0000000000000000 R15: 0000000000000000 [ 699.729687] ---[ end trace 4ce02f25ff7d3df5 ]--- [ 699.729782] ------------[ cut here ]------------ [ 699.729785] kernel BUG at fs/f2fs/segment.h:654! [ 699.731055] invalid opcode: 0000 [#1] SMP KASAN PTI [ 699.732104] CPU: 0 PID: 1309 Comm: a.out Tainted: G W 4.18.0-rc1+ #4 [ 699.733684] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014 [ 699.735611] RIP: 0010:f2fs_submit_page_bio+0x29b/0x730 [ 699.736649] Code: 54 49 8d bd 18 04 00 00 e8 b2 59 af ff 41 8b 8d 18 04 00 00 8b 45 b8 41 d3 e6 44 01 f0 4c 8d 73 14 41 39 c7 0f 82 37 fe ff ff <0f> 0b 65 8b 05 2c 04 77 47 89 c0 48 0f a3 05 52 c1 d5 01 0f 92 c0 [ 699.740524] RSP: 0018:ffff8801f43af508 EFLAGS: 00010283 [ 699.741573] RAX: 0000000000000000 RBX: ffff8801f43af7b8 RCX: ffffffffb88a7cef [ 699.743006] RDX: 0000000000000007 RSI: dffffc0000000000 RDI: ffff8801e3e7a64c [ 699.744426] RBP: ffff8801f43af558 R08: ffffed003e066b55 R09: ffffed003e066b55 [ 699.745833] R10: 0000000000000001 R11: ffffed003e066b54 R12: ffffea0007876940 [ 699.747256] R13: ffff8801f0335500 R14: ffff8801e3e7a600 R15: 0000000000000001 [ 699.748683] FS: 00007f9bf97f5700(0000) GS:ffff8801f6e00000(0000) knlGS:0000000000000000 [ 699.750293] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 699.751462] CR2: 00007f9bf925d170 CR3: 00000001f0c34000 CR4: 00000000000006f0 [ 699.752874] Call Trace: [ 699.753386] ? f2fs_inplace_write_data+0x93/0x240 [ 699.754341] f2fs_inplace_write_data+0xd2/0x240 [ 699.755271] f2fs_do_write_data_page+0x2e2/0xe00 [ 699.756214] ? f2fs_should_update_outplace+0xd0/0xd0 [ 699.757215] ? memcg_drain_all_list_lrus+0x280/0x280 [ 699.758209] ? __radix_tree_replace+0xa3/0x120 [ 699.759164] __write_data_page+0x5c7/0xe30 [ 699.760002] ? kasan_check_read+0x11/0x20 [ 699.760823] ? page_mapped+0x8a/0x110 [ 699.761573] ? page_mkclean+0xe9/0x160 [ 699.762345] ? f2fs_do_write_data_page+0xe00/0xe00 [ 699.763332] ? invalid_page_referenced_vma+0x130/0x130 [ 699.764374] ? clear_page_dirty_for_io+0x332/0x450 [ 699.765347] f2fs_write_cache_pages+0x4ca/0x860 [ 699.766276] ? __write_data_page+0xe30/0xe30 [ 699.767161] ? percpu_counter_add_batch+0x22/0xa0 [ 699.768112] ? kasan_check_write+0x14/0x20 [ 699.768951] ? _raw_spin_lock+0x17/0x40 [ 699.769739] ? f2fs_mark_inode_dirty_sync.part.18+0x16/0x30 [ 699.770885] ? iov_iter_advance+0x113/0x640 [ 699.771743] ? f2fs_write_end+0x133/0x2e0 [ 699.772569] ? balance_dirty_pages_ratelimited+0x239/0x640 [ 699.773680] f2fs_write_data_pages+0x329/0x520 [ 699.774603] ? generic_perform_write+0x250/0x320 [ 699.775544] ? f2fs_write_cache_pages+0x860/0x860 [ 699.776510] ? current_time+0x110/0x110 [ 699.777299] ? f2fs_preallocate_blocks+0x1ef/0x370 [ 699.778279] do_writepages+0x37/0xb0 [ 699.779026] ? f2fs_write_cache_pages+0x860/0x860 [ 699.779978] ? do_writepages+0x37/0xb0 [ 699.780755] __filemap_fdatawrite_range+0x19a/0x1f0 [ 699.781746] ? delete_from_page_cache_batch+0x4e0/0x4e0 [ 699.782820] ? __vfs_write+0x2b2/0x410 [ 699.783597] file_write_and_wait_range+0x66/0xb0 [ 699.784540] f2fs_do_sync_file+0x1f9/0xd90 [ 699.785381] ? truncate_partial_data_page+0x290/0x290 [ 699.786415] ? __sb_end_write+0x30/0x50 [ 699.787204] ? vfs_write+0x20f/0x260 [ 699.787941] f2fs_sync_file+0x9a/0xb0 [ 699.788694] ? f2fs_do_sync_file+0xd90/0xd90 [ 699.789572] vfs_fsync_range+0x68/0x100 [ 699.790360] ? __fget_light+0xc9/0xe0 [ 699.791128] do_fsync+0x3d/0x70 [ 699.791779] __x64_sys_fdatasync+0x24/0x30 [ 699.792614] do_syscall_64+0x78/0x170 [ 699.793371] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 699.794406] RIP: 0033:0x7f9bf930d800 [ 699.795134] Code: 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 83 3d 49 bf 2c 00 00 75 10 b8 4b 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 31 c3 48 83 ec 08 e8 be 78 01 00 48 89 04 24 [ 699.798960] RSP: 002b:00007ffee3606c68 EFLAGS: 00000246 ORIG_RAX: 000000000000004b [ 699.800483] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f9bf930d800 [ 699.801923] RDX: 0000000000008000 RSI: 00000000006010a0 RDI: 0000000000000003 [ 699.803373] RBP: 00007ffee3606ca0 R08: 0000000001503010 R09: 0000000000000000 [ 699.804798] R10: 00000000000002e8 R11: 0000000000000246 R12: 0000000000400610 [ 699.806233] R13: 00007ffee3606da0 R14: 0000000000000000 R15: 0000000000000000 [ 699.807667] Modules linked in: snd_hda_codec_generic snd_hda_intel snd_hda_codec snd_hwdep snd_hda_core snd_pcm snd_timer snd mac_hid i2c_piix4 soundcore ib_iser rdma_cm iw_cm ib_cm ib_core iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx raid1 raid0 multipath linear 8139too crct10dif_pclmul crc32_pclmul qxl drm_kms_helper syscopyarea aesni_intel sysfillrect sysimgblt fb_sys_fops ttm drm aes_x86_64 crypto_simd cryptd 8139cp glue_helper mii pata_acpi floppy [ 699.817079] ---[ end trace 4ce02f25ff7d3df6 ]--- [ 699.818068] RIP: 0010:f2fs_submit_page_bio+0x29b/0x730 [ 699.819114] Code: 54 49 8d bd 18 04 00 00 e8 b2 59 af ff 41 8b 8d 18 04 00 00 8b 45 b8 41 d3 e6 44 01 f0 4c 8d 73 14 41 39 c7 0f 82 37 fe ff ff <0f> 0b 65 8b 05 2c 04 77 47 89 c0 48 0f a3 05 52 c1 d5 01 0f 92 c0 [ 699.822919] RSP: 0018:ffff8801f43af508 EFLAGS: 00010283 [ 699.823977] RAX: 0000000000000000 RBX: ffff8801f43af7b8 RCX: ffffffffb88a7cef [ 699.825436] RDX: 0000000000000007 RSI: dffffc0000000000 RDI: ffff8801e3e7a64c [ 699.826881] RBP: ffff8801f43af558 R08: ffffed003e066b55 R09: ffffed003e066b55 [ 699.828292] R10: 0000000000000001 R11: ffffed003e066b54 R12: ffffea0007876940 [ 699.829750] R13: ffff8801f0335500 R14: ffff8801e3e7a600 R15: 0000000000000001 [ 699.831192] FS: 00007f9bf97f5700(0000) GS:ffff8801f6e00000(0000) knlGS:0000000000000000 [ 699.832793] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 699.833981] CR2: 00007f9bf925d170 CR3: 00000001f0c34000 CR4: 00000000000006f0 [ 699.835556] ================================================================== [ 699.837029] BUG: KASAN: stack-out-of-bounds in update_stack_state+0x38c/0x3e0 [ 699.838462] Read of size 8 at addr ffff8801f43af970 by task a.out/1309 [ 699.840086] CPU: 0 PID: 1309 Comm: a.out Tainted: G D W 4.18.0-rc1+ #4 [ 699.841603] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014 [ 699.843475] Call Trace: [ 699.843982] dump_stack+0x7b/0xb5 [ 699.844661] print_address_description+0x70/0x290 [ 699.845607] kasan_report+0x291/0x390 [ 699.846351] ? update_stack_state+0x38c/0x3e0 [ 699.853831] __asan_load8+0x54/0x90 [ 699.854569] update_stack_state+0x38c/0x3e0 [ 699.855428] ? __read_once_size_nocheck.constprop.7+0x20/0x20 [ 699.856601] ? __save_stack_trace+0x5e/0x100 [ 699.857476] unwind_next_frame.part.5+0x18e/0x490 [ 699.858448] ? unwind_dump+0x290/0x290 [ 699.859217] ? clear_page_dirty_for_io+0x332/0x450 [ 699.860185] __unwind_start+0x106/0x190 [ 699.860974] __save_stack_trace+0x5e/0x100 [ 699.861808] ? __save_stack_trace+0x5e/0x100 [ 699.862691] ? unlink_anon_vmas+0xba/0x2c0 [ 699.863525] save_stack_trace+0x1f/0x30 [ 699.864312] save_stack+0x46/0xd0 [ 699.864993] ? __alloc_pages_slowpath+0x1420/0x1420 [ 699.865990] ? flush_tlb_mm_range+0x15e/0x220 [ 699.866889] ? kasan_check_write+0x14/0x20 [ 699.867724] ? __dec_node_state+0x92/0xb0 [ 699.868543] ? lock_page_memcg+0x85/0xf0 [ 699.869350] ? unlock_page_memcg+0x16/0x80 [ 699.870185] ? page_remove_rmap+0x198/0x520 [ 699.871048] ? mark_page_accessed+0x133/0x200 [ 699.871930] ? _cond_resched+0x1a/0x50 [ 699.872700] ? unmap_page_range+0xcd4/0xe50 [ 699.873551] ? rb_next+0x58/0x80 [ 699.874217] ? rb_next+0x58/0x80 [ 699.874895] __kasan_slab_free+0x13c/0x1a0 [ 699.875734] ? unlink_anon_vmas+0xba/0x2c0 [ 699.876563] kasan_slab_free+0xe/0x10 [ 699.877315] kmem_cache_free+0x89/0x1e0 [ 699.878095] unlink_anon_vmas+0xba/0x2c0 [ 699.878913] free_pgtables+0x101/0x1b0 [ 699.879677] exit_mmap+0x146/0x2a0 [ 699.880378] ? __ia32_sys_munmap+0x50/0x50 [ 699.881214] ? kasan_check_read+0x11/0x20 [ 699.882052] ? mm_update_next_owner+0x322/0x380 [ 699.882985] mmput+0x8b/0x1d0 [ 699.883602] do_exit+0x43a/0x1390 [ 699.884288] ? mm_update_next_owner+0x380/0x380 [ 699.885212] ? f2fs_sync_file+0x9a/0xb0 [ 699.885995] ? f2fs_do_sync_file+0xd90/0xd90 [ 699.886877] ? vfs_fsync_range+0x68/0x100 [ 699.887694] ? __fget_light+0xc9/0xe0 [ 699.888442] ? do_fsync+0x3d/0x70 [ 699.889118] ? __x64_sys_fdatasync+0x24/0x30 [ 699.889996] rewind_stack_do_exit+0x17/0x20 [ 699.890860] RIP: 0033:0x7f9bf930d800 [ 699.891585] Code: Bad RIP value. [ 699.892268] RSP: 002b:00007ffee3606c68 EFLAGS: 00000246 ORIG_RAX: 000000000000004b [ 699.893781] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f9bf930d800 [ 699.895220] RDX: 0000000000008000 RSI: 00000000006010a0 RDI: 0000000000000003 [ 699.896643] RBP: 00007ffee3606ca0 R08: 0000000001503010 R09: 0000000000000000 [ 699.898069] R10: 00000000000002e8 R11: 0000000000000246 R12: 0000000000400610 [ 699.899505] R13: 00007ffee3606da0 R14: 0000000000000000 R15: 0000000000000000 [ 699.901241] The buggy address belongs to the page: [ 699.902215] page:ffffea0007d0ebc0 count:0 mapcount:0 mapping:0000000000000000 index:0x0 [ 699.903811] flags: 0x2ffff0000000000() [ 699.904585] raw: 02ffff0000000000 0000000000000000 ffffffff07d00101 0000000000000000 [ 699.906125] raw: 0000000000000000 0000000000240000 00000000ffffffff 0000000000000000 [ 699.907673] page dumped because: kasan: bad access detected [ 699.909108] Memory state around the buggy address: [ 699.910077] ffff8801f43af800: 00 f1 f1 f1 f1 00 f4 f4 f4 f3 f3 f3 f3 00 00 00 [ 699.911528] ffff8801f43af880: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 [ 699.912953] >ffff8801f43af900: 00 00 00 00 00 00 00 00 f1 01 f4 f4 f4 f2 f2 f2 [ 699.914392] ^ [ 699.915758] ffff8801f43af980: f2 00 f4 f4 00 00 00 00 f2 00 00 00 00 00 00 00 [ 699.917193] ffff8801f43afa00: 00 00 00 00 00 00 00 00 00 f3 f3 f3 00 00 00 00 [ 699.918634] ================================================================== - Location https://elixir.bootlin.com/linux/v4.18-rc1/source/fs/f2fs/segment.h#L644 Reported-by Wen Xu <wen.xu@gatech.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-01 11:13:44 +00:00
goto redirty_out;
}
f2fs: fix to do sanity check with block address in main area This patch add to do sanity check with below field: - cp_pack_total_block_count - blkaddr of data/node - extent info - Overview BUG() in verify_block_addr() when writing to a corrupted f2fs image - Reproduce (4.18 upstream kernel) - POC (poc.c) static void activity(char *mpoint) { char *foo_bar_baz; int err; static int buf[8192]; memset(buf, 0, sizeof(buf)); err = asprintf(&foo_bar_baz, "%s/foo/bar/baz", mpoint); int fd = open(foo_bar_baz, O_RDWR | O_TRUNC, 0777); if (fd >= 0) { write(fd, (char *)buf, sizeof(buf)); fdatasync(fd); close(fd); } } int main(int argc, char *argv[]) { activity(argv[1]); return 0; } - Kernel message [ 689.349473] F2FS-fs (loop0): Mounted with checkpoint version = 3 [ 699.728662] WARNING: CPU: 0 PID: 1309 at fs/f2fs/segment.c:2860 f2fs_inplace_write_data+0x232/0x240 [ 699.728670] Modules linked in: snd_hda_codec_generic snd_hda_intel snd_hda_codec snd_hwdep snd_hda_core snd_pcm snd_timer snd mac_hid i2c_piix4 soundcore ib_iser rdma_cm iw_cm ib_cm ib_core iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx raid1 raid0 multipath linear 8139too crct10dif_pclmul crc32_pclmul qxl drm_kms_helper syscopyarea aesni_intel sysfillrect sysimgblt fb_sys_fops ttm drm aes_x86_64 crypto_simd cryptd 8139cp glue_helper mii pata_acpi floppy [ 699.729056] CPU: 0 PID: 1309 Comm: a.out Not tainted 4.18.0-rc1+ #4 [ 699.729064] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014 [ 699.729074] RIP: 0010:f2fs_inplace_write_data+0x232/0x240 [ 699.729076] Code: ff e9 cf fe ff ff 49 8d 7d 10 e8 39 45 ad ff 4d 8b 7d 10 be 04 00 00 00 49 8d 7f 48 e8 07 49 ad ff 45 8b 7f 48 e9 fb fe ff ff <0f> 0b f0 41 80 4d 48 04 e9 65 fe ff ff 90 66 66 66 66 90 55 48 8d [ 699.729130] RSP: 0018:ffff8801f43af568 EFLAGS: 00010202 [ 699.729139] RAX: 000000000000003f RBX: ffff8801f43af7b8 RCX: ffffffffb88c9113 [ 699.729142] RDX: 0000000000000003 RSI: dffffc0000000000 RDI: ffff8802024e5540 [ 699.729144] RBP: ffff8801f43af590 R08: 0000000000000009 R09: ffffffffffffffe8 [ 699.729147] R10: 0000000000000001 R11: ffffed0039b0596a R12: ffff8802024e5540 [ 699.729149] R13: ffff8801f0335500 R14: ffff8801e3e7a700 R15: ffff8801e1ee4450 [ 699.729154] FS: 00007f9bf97f5700(0000) GS:ffff8801f6e00000(0000) knlGS:0000000000000000 [ 699.729156] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 699.729159] CR2: 00007f9bf925d170 CR3: 00000001f0c34000 CR4: 00000000000006f0 [ 699.729171] Call Trace: [ 699.729192] f2fs_do_write_data_page+0x2e2/0xe00 [ 699.729203] ? f2fs_should_update_outplace+0xd0/0xd0 [ 699.729238] ? memcg_drain_all_list_lrus+0x280/0x280 [ 699.729269] ? __radix_tree_replace+0xa3/0x120 [ 699.729276] __write_data_page+0x5c7/0xe30 [ 699.729291] ? kasan_check_read+0x11/0x20 [ 699.729310] ? page_mapped+0x8a/0x110 [ 699.729321] ? page_mkclean+0xe9/0x160 [ 699.729327] ? f2fs_do_write_data_page+0xe00/0xe00 [ 699.729331] ? invalid_page_referenced_vma+0x130/0x130 [ 699.729345] ? clear_page_dirty_for_io+0x332/0x450 [ 699.729351] f2fs_write_cache_pages+0x4ca/0x860 [ 699.729358] ? __write_data_page+0xe30/0xe30 [ 699.729374] ? percpu_counter_add_batch+0x22/0xa0 [ 699.729380] ? kasan_check_write+0x14/0x20 [ 699.729391] ? _raw_spin_lock+0x17/0x40 [ 699.729403] ? f2fs_mark_inode_dirty_sync.part.18+0x16/0x30 [ 699.729413] ? iov_iter_advance+0x113/0x640 [ 699.729418] ? f2fs_write_end+0x133/0x2e0 [ 699.729423] ? balance_dirty_pages_ratelimited+0x239/0x640 [ 699.729428] f2fs_write_data_pages+0x329/0x520 [ 699.729433] ? generic_perform_write+0x250/0x320 [ 699.729438] ? f2fs_write_cache_pages+0x860/0x860 [ 699.729454] ? current_time+0x110/0x110 [ 699.729459] ? f2fs_preallocate_blocks+0x1ef/0x370 [ 699.729464] do_writepages+0x37/0xb0 [ 699.729468] ? f2fs_write_cache_pages+0x860/0x860 [ 699.729472] ? do_writepages+0x37/0xb0 [ 699.729478] __filemap_fdatawrite_range+0x19a/0x1f0 [ 699.729483] ? delete_from_page_cache_batch+0x4e0/0x4e0 [ 699.729496] ? __vfs_write+0x2b2/0x410 [ 699.729501] file_write_and_wait_range+0x66/0xb0 [ 699.729506] f2fs_do_sync_file+0x1f9/0xd90 [ 699.729511] ? truncate_partial_data_page+0x290/0x290 [ 699.729521] ? __sb_end_write+0x30/0x50 [ 699.729526] ? vfs_write+0x20f/0x260 [ 699.729530] f2fs_sync_file+0x9a/0xb0 [ 699.729534] ? f2fs_do_sync_file+0xd90/0xd90 [ 699.729548] vfs_fsync_range+0x68/0x100 [ 699.729554] ? __fget_light+0xc9/0xe0 [ 699.729558] do_fsync+0x3d/0x70 [ 699.729562] __x64_sys_fdatasync+0x24/0x30 [ 699.729585] do_syscall_64+0x78/0x170 [ 699.729595] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 699.729613] RIP: 0033:0x7f9bf930d800 [ 699.729615] Code: 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 83 3d 49 bf 2c 00 00 75 10 b8 4b 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 31 c3 48 83 ec 08 e8 be 78 01 00 48 89 04 24 [ 699.729668] RSP: 002b:00007ffee3606c68 EFLAGS: 00000246 ORIG_RAX: 000000000000004b [ 699.729673] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f9bf930d800 [ 699.729675] RDX: 0000000000008000 RSI: 00000000006010a0 RDI: 0000000000000003 [ 699.729678] RBP: 00007ffee3606ca0 R08: 0000000001503010 R09: 0000000000000000 [ 699.729680] R10: 00000000000002e8 R11: 0000000000000246 R12: 0000000000400610 [ 699.729683] R13: 00007ffee3606da0 R14: 0000000000000000 R15: 0000000000000000 [ 699.729687] ---[ end trace 4ce02f25ff7d3df5 ]--- [ 699.729782] ------------[ cut here ]------------ [ 699.729785] kernel BUG at fs/f2fs/segment.h:654! [ 699.731055] invalid opcode: 0000 [#1] SMP KASAN PTI [ 699.732104] CPU: 0 PID: 1309 Comm: a.out Tainted: G W 4.18.0-rc1+ #4 [ 699.733684] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014 [ 699.735611] RIP: 0010:f2fs_submit_page_bio+0x29b/0x730 [ 699.736649] Code: 54 49 8d bd 18 04 00 00 e8 b2 59 af ff 41 8b 8d 18 04 00 00 8b 45 b8 41 d3 e6 44 01 f0 4c 8d 73 14 41 39 c7 0f 82 37 fe ff ff <0f> 0b 65 8b 05 2c 04 77 47 89 c0 48 0f a3 05 52 c1 d5 01 0f 92 c0 [ 699.740524] RSP: 0018:ffff8801f43af508 EFLAGS: 00010283 [ 699.741573] RAX: 0000000000000000 RBX: ffff8801f43af7b8 RCX: ffffffffb88a7cef [ 699.743006] RDX: 0000000000000007 RSI: dffffc0000000000 RDI: ffff8801e3e7a64c [ 699.744426] RBP: ffff8801f43af558 R08: ffffed003e066b55 R09: ffffed003e066b55 [ 699.745833] R10: 0000000000000001 R11: ffffed003e066b54 R12: ffffea0007876940 [ 699.747256] R13: ffff8801f0335500 R14: ffff8801e3e7a600 R15: 0000000000000001 [ 699.748683] FS: 00007f9bf97f5700(0000) GS:ffff8801f6e00000(0000) knlGS:0000000000000000 [ 699.750293] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 699.751462] CR2: 00007f9bf925d170 CR3: 00000001f0c34000 CR4: 00000000000006f0 [ 699.752874] Call Trace: [ 699.753386] ? f2fs_inplace_write_data+0x93/0x240 [ 699.754341] f2fs_inplace_write_data+0xd2/0x240 [ 699.755271] f2fs_do_write_data_page+0x2e2/0xe00 [ 699.756214] ? f2fs_should_update_outplace+0xd0/0xd0 [ 699.757215] ? memcg_drain_all_list_lrus+0x280/0x280 [ 699.758209] ? __radix_tree_replace+0xa3/0x120 [ 699.759164] __write_data_page+0x5c7/0xe30 [ 699.760002] ? kasan_check_read+0x11/0x20 [ 699.760823] ? page_mapped+0x8a/0x110 [ 699.761573] ? page_mkclean+0xe9/0x160 [ 699.762345] ? f2fs_do_write_data_page+0xe00/0xe00 [ 699.763332] ? invalid_page_referenced_vma+0x130/0x130 [ 699.764374] ? clear_page_dirty_for_io+0x332/0x450 [ 699.765347] f2fs_write_cache_pages+0x4ca/0x860 [ 699.766276] ? __write_data_page+0xe30/0xe30 [ 699.767161] ? percpu_counter_add_batch+0x22/0xa0 [ 699.768112] ? kasan_check_write+0x14/0x20 [ 699.768951] ? _raw_spin_lock+0x17/0x40 [ 699.769739] ? f2fs_mark_inode_dirty_sync.part.18+0x16/0x30 [ 699.770885] ? iov_iter_advance+0x113/0x640 [ 699.771743] ? f2fs_write_end+0x133/0x2e0 [ 699.772569] ? balance_dirty_pages_ratelimited+0x239/0x640 [ 699.773680] f2fs_write_data_pages+0x329/0x520 [ 699.774603] ? generic_perform_write+0x250/0x320 [ 699.775544] ? f2fs_write_cache_pages+0x860/0x860 [ 699.776510] ? current_time+0x110/0x110 [ 699.777299] ? f2fs_preallocate_blocks+0x1ef/0x370 [ 699.778279] do_writepages+0x37/0xb0 [ 699.779026] ? f2fs_write_cache_pages+0x860/0x860 [ 699.779978] ? do_writepages+0x37/0xb0 [ 699.780755] __filemap_fdatawrite_range+0x19a/0x1f0 [ 699.781746] ? delete_from_page_cache_batch+0x4e0/0x4e0 [ 699.782820] ? __vfs_write+0x2b2/0x410 [ 699.783597] file_write_and_wait_range+0x66/0xb0 [ 699.784540] f2fs_do_sync_file+0x1f9/0xd90 [ 699.785381] ? truncate_partial_data_page+0x290/0x290 [ 699.786415] ? __sb_end_write+0x30/0x50 [ 699.787204] ? vfs_write+0x20f/0x260 [ 699.787941] f2fs_sync_file+0x9a/0xb0 [ 699.788694] ? f2fs_do_sync_file+0xd90/0xd90 [ 699.789572] vfs_fsync_range+0x68/0x100 [ 699.790360] ? __fget_light+0xc9/0xe0 [ 699.791128] do_fsync+0x3d/0x70 [ 699.791779] __x64_sys_fdatasync+0x24/0x30 [ 699.792614] do_syscall_64+0x78/0x170 [ 699.793371] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 699.794406] RIP: 0033:0x7f9bf930d800 [ 699.795134] Code: 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 83 3d 49 bf 2c 00 00 75 10 b8 4b 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 31 c3 48 83 ec 08 e8 be 78 01 00 48 89 04 24 [ 699.798960] RSP: 002b:00007ffee3606c68 EFLAGS: 00000246 ORIG_RAX: 000000000000004b [ 699.800483] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f9bf930d800 [ 699.801923] RDX: 0000000000008000 RSI: 00000000006010a0 RDI: 0000000000000003 [ 699.803373] RBP: 00007ffee3606ca0 R08: 0000000001503010 R09: 0000000000000000 [ 699.804798] R10: 00000000000002e8 R11: 0000000000000246 R12: 0000000000400610 [ 699.806233] R13: 00007ffee3606da0 R14: 0000000000000000 R15: 0000000000000000 [ 699.807667] Modules linked in: snd_hda_codec_generic snd_hda_intel snd_hda_codec snd_hwdep snd_hda_core snd_pcm snd_timer snd mac_hid i2c_piix4 soundcore ib_iser rdma_cm iw_cm ib_cm ib_core iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx raid1 raid0 multipath linear 8139too crct10dif_pclmul crc32_pclmul qxl drm_kms_helper syscopyarea aesni_intel sysfillrect sysimgblt fb_sys_fops ttm drm aes_x86_64 crypto_simd cryptd 8139cp glue_helper mii pata_acpi floppy [ 699.817079] ---[ end trace 4ce02f25ff7d3df6 ]--- [ 699.818068] RIP: 0010:f2fs_submit_page_bio+0x29b/0x730 [ 699.819114] Code: 54 49 8d bd 18 04 00 00 e8 b2 59 af ff 41 8b 8d 18 04 00 00 8b 45 b8 41 d3 e6 44 01 f0 4c 8d 73 14 41 39 c7 0f 82 37 fe ff ff <0f> 0b 65 8b 05 2c 04 77 47 89 c0 48 0f a3 05 52 c1 d5 01 0f 92 c0 [ 699.822919] RSP: 0018:ffff8801f43af508 EFLAGS: 00010283 [ 699.823977] RAX: 0000000000000000 RBX: ffff8801f43af7b8 RCX: ffffffffb88a7cef [ 699.825436] RDX: 0000000000000007 RSI: dffffc0000000000 RDI: ffff8801e3e7a64c [ 699.826881] RBP: ffff8801f43af558 R08: ffffed003e066b55 R09: ffffed003e066b55 [ 699.828292] R10: 0000000000000001 R11: ffffed003e066b54 R12: ffffea0007876940 [ 699.829750] R13: ffff8801f0335500 R14: ffff8801e3e7a600 R15: 0000000000000001 [ 699.831192] FS: 00007f9bf97f5700(0000) GS:ffff8801f6e00000(0000) knlGS:0000000000000000 [ 699.832793] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 699.833981] CR2: 00007f9bf925d170 CR3: 00000001f0c34000 CR4: 00000000000006f0 [ 699.835556] ================================================================== [ 699.837029] BUG: KASAN: stack-out-of-bounds in update_stack_state+0x38c/0x3e0 [ 699.838462] Read of size 8 at addr ffff8801f43af970 by task a.out/1309 [ 699.840086] CPU: 0 PID: 1309 Comm: a.out Tainted: G D W 4.18.0-rc1+ #4 [ 699.841603] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014 [ 699.843475] Call Trace: [ 699.843982] dump_stack+0x7b/0xb5 [ 699.844661] print_address_description+0x70/0x290 [ 699.845607] kasan_report+0x291/0x390 [ 699.846351] ? update_stack_state+0x38c/0x3e0 [ 699.853831] __asan_load8+0x54/0x90 [ 699.854569] update_stack_state+0x38c/0x3e0 [ 699.855428] ? __read_once_size_nocheck.constprop.7+0x20/0x20 [ 699.856601] ? __save_stack_trace+0x5e/0x100 [ 699.857476] unwind_next_frame.part.5+0x18e/0x490 [ 699.858448] ? unwind_dump+0x290/0x290 [ 699.859217] ? clear_page_dirty_for_io+0x332/0x450 [ 699.860185] __unwind_start+0x106/0x190 [ 699.860974] __save_stack_trace+0x5e/0x100 [ 699.861808] ? __save_stack_trace+0x5e/0x100 [ 699.862691] ? unlink_anon_vmas+0xba/0x2c0 [ 699.863525] save_stack_trace+0x1f/0x30 [ 699.864312] save_stack+0x46/0xd0 [ 699.864993] ? __alloc_pages_slowpath+0x1420/0x1420 [ 699.865990] ? flush_tlb_mm_range+0x15e/0x220 [ 699.866889] ? kasan_check_write+0x14/0x20 [ 699.867724] ? __dec_node_state+0x92/0xb0 [ 699.868543] ? lock_page_memcg+0x85/0xf0 [ 699.869350] ? unlock_page_memcg+0x16/0x80 [ 699.870185] ? page_remove_rmap+0x198/0x520 [ 699.871048] ? mark_page_accessed+0x133/0x200 [ 699.871930] ? _cond_resched+0x1a/0x50 [ 699.872700] ? unmap_page_range+0xcd4/0xe50 [ 699.873551] ? rb_next+0x58/0x80 [ 699.874217] ? rb_next+0x58/0x80 [ 699.874895] __kasan_slab_free+0x13c/0x1a0 [ 699.875734] ? unlink_anon_vmas+0xba/0x2c0 [ 699.876563] kasan_slab_free+0xe/0x10 [ 699.877315] kmem_cache_free+0x89/0x1e0 [ 699.878095] unlink_anon_vmas+0xba/0x2c0 [ 699.878913] free_pgtables+0x101/0x1b0 [ 699.879677] exit_mmap+0x146/0x2a0 [ 699.880378] ? __ia32_sys_munmap+0x50/0x50 [ 699.881214] ? kasan_check_read+0x11/0x20 [ 699.882052] ? mm_update_next_owner+0x322/0x380 [ 699.882985] mmput+0x8b/0x1d0 [ 699.883602] do_exit+0x43a/0x1390 [ 699.884288] ? mm_update_next_owner+0x380/0x380 [ 699.885212] ? f2fs_sync_file+0x9a/0xb0 [ 699.885995] ? f2fs_do_sync_file+0xd90/0xd90 [ 699.886877] ? vfs_fsync_range+0x68/0x100 [ 699.887694] ? __fget_light+0xc9/0xe0 [ 699.888442] ? do_fsync+0x3d/0x70 [ 699.889118] ? __x64_sys_fdatasync+0x24/0x30 [ 699.889996] rewind_stack_do_exit+0x17/0x20 [ 699.890860] RIP: 0033:0x7f9bf930d800 [ 699.891585] Code: Bad RIP value. [ 699.892268] RSP: 002b:00007ffee3606c68 EFLAGS: 00000246 ORIG_RAX: 000000000000004b [ 699.893781] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f9bf930d800 [ 699.895220] RDX: 0000000000008000 RSI: 00000000006010a0 RDI: 0000000000000003 [ 699.896643] RBP: 00007ffee3606ca0 R08: 0000000001503010 R09: 0000000000000000 [ 699.898069] R10: 00000000000002e8 R11: 0000000000000246 R12: 0000000000400610 [ 699.899505] R13: 00007ffee3606da0 R14: 0000000000000000 R15: 0000000000000000 [ 699.901241] The buggy address belongs to the page: [ 699.902215] page:ffffea0007d0ebc0 count:0 mapcount:0 mapping:0000000000000000 index:0x0 [ 699.903811] flags: 0x2ffff0000000000() [ 699.904585] raw: 02ffff0000000000 0000000000000000 ffffffff07d00101 0000000000000000 [ 699.906125] raw: 0000000000000000 0000000000240000 00000000ffffffff 0000000000000000 [ 699.907673] page dumped because: kasan: bad access detected [ 699.909108] Memory state around the buggy address: [ 699.910077] ffff8801f43af800: 00 f1 f1 f1 f1 00 f4 f4 f4 f3 f3 f3 f3 00 00 00 [ 699.911528] ffff8801f43af880: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 [ 699.912953] >ffff8801f43af900: 00 00 00 00 00 00 00 00 f1 01 f4 f4 f4 f2 f2 f2 [ 699.914392] ^ [ 699.915758] ffff8801f43af980: f2 00 f4 f4 00 00 00 00 f2 00 00 00 00 00 00 00 [ 699.917193] ffff8801f43afa00: 00 00 00 00 00 00 00 00 00 f3 f3 f3 00 00 00 00 [ 699.918634] ================================================================== - Location https://elixir.bootlin.com/linux/v4.18-rc1/source/fs/f2fs/segment.h#L644 Reported-by Wen Xu <wen.xu@gatech.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-01 11:13:44 +00:00
if (atomic && !test_opt(sbi, NOBARRIER))
fio.op_flags |= REQ_PREFLUSH | REQ_FUA;
set_page_writeback(page);
ClearPageError(page);
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
if (f2fs_in_warm_node_list(sbi, page)) {
seq = f2fs_add_fsync_node_entry(sbi, page);
if (seq_id)
*seq_id = seq;
}
fio.old_blkaddr = ni.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_do_write_node_page(nid, &fio);
set_node_addr(sbi, &ni, fio.new_blkaddr, is_fsync_dnode(page));
dec_page_count(sbi, F2FS_DIRTY_NODES);
up_read(&sbi->node_write);
if (wbc->for_reclaim) {
f2fs_submit_merged_write_cond(sbi, NULL, page, 0, NODE);
submitted = NULL;
}
unlock_page(page);
if (unlikely(f2fs_cp_error(sbi))) {
f2fs_submit_merged_write(sbi, NODE);
submitted = NULL;
}
if (submitted)
*submitted = fio.submitted;
if (do_balance)
f2fs_balance_fs(sbi, false);
return 0;
redirty_out:
redirty_page_for_writepage(wbc, page);
return AOP_WRITEPAGE_ACTIVATE;
}
int f2fs_move_node_page(struct page *node_page, int gc_type)
{
int err = 0;
if (gc_type == FG_GC) {
struct writeback_control wbc = {
.sync_mode = WB_SYNC_ALL,
.nr_to_write = 1,
.for_reclaim = 0,
};
f2fs_wait_on_page_writeback(node_page, NODE, true, true);
set_page_dirty(node_page);
if (!clear_page_dirty_for_io(node_page)) {
err = -EAGAIN;
goto out_page;
}
if (__write_node_page(node_page, false, NULL,
&wbc, false, FS_GC_NODE_IO, NULL)) {
err = -EAGAIN;
unlock_page(node_page);
}
goto release_page;
} else {
/* set page dirty and write it */
if (!PageWriteback(node_page))
set_page_dirty(node_page);
}
out_page:
unlock_page(node_page);
release_page:
f2fs_put_page(node_page, 0);
return err;
}
static int f2fs_write_node_page(struct page *page,
struct writeback_control *wbc)
{
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
return __write_node_page(page, false, NULL, wbc, false,
FS_NODE_IO, 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
int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
struct writeback_control *wbc, bool atomic,
unsigned int *seq_id)
{
pgoff_t index;
struct pagevec pvec;
int ret = 0;
struct page *last_page = NULL;
bool marked = false;
nid_t ino = inode->i_ino;
int nr_pages;
int nwritten = 0;
if (atomic) {
last_page = last_fsync_dnode(sbi, ino);
if (IS_ERR_OR_NULL(last_page))
return PTR_ERR_OR_ZERO(last_page);
}
retry:
pagevec_init(&pvec);
index = 0;
while ((nr_pages = pagevec_lookup_tag(&pvec, NODE_MAPPING(sbi), &index,
PAGECACHE_TAG_DIRTY))) {
int i;
for (i = 0; i < nr_pages; i++) {
struct page *page = pvec.pages[i];
bool submitted = false;
if (unlikely(f2fs_cp_error(sbi))) {
f2fs_put_page(last_page, 0);
pagevec_release(&pvec);
ret = -EIO;
goto out;
}
if (!IS_DNODE(page) || !is_cold_node(page))
continue;
if (ino_of_node(page) != ino)
continue;
lock_page(page);
if (unlikely(page->mapping != NODE_MAPPING(sbi))) {
continue_unlock:
unlock_page(page);
continue;
}
if (ino_of_node(page) != ino)
goto continue_unlock;
if (!PageDirty(page) && page != last_page) {
/* someone wrote it for us */
goto continue_unlock;
}
f2fs_wait_on_page_writeback(page, NODE, true, true);
set_fsync_mark(page, 0);
set_dentry_mark(page, 0);
if (!atomic || page == last_page) {
set_fsync_mark(page, 1);
if (IS_INODE(page)) {
if (is_inode_flag_set(inode,
FI_DIRTY_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_update_inode(inode, page);
set_dentry_mark(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
f2fs_need_dentry_mark(sbi, ino));
}
/* may be written by other thread */
if (!PageDirty(page))
set_page_dirty(page);
}
if (!clear_page_dirty_for_io(page))
goto continue_unlock;
ret = __write_node_page(page, atomic &&
page == last_page,
&submitted, wbc, true,
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
FS_NODE_IO, seq_id);
if (ret) {
unlock_page(page);
f2fs_put_page(last_page, 0);
break;
} else if (submitted) {
nwritten++;
}
if (page == last_page) {
f2fs_put_page(page, 0);
marked = true;
break;
}
}
pagevec_release(&pvec);
cond_resched();
if (ret || marked)
break;
}
if (!ret && atomic && !marked) {
f2fs_debug(sbi, "Retry to write fsync mark: ino=%u, idx=%lx",
ino, last_page->index);
lock_page(last_page);
f2fs_wait_on_page_writeback(last_page, NODE, true, true);
set_page_dirty(last_page);
unlock_page(last_page);
goto retry;
}
out:
if (nwritten)
f2fs_submit_merged_write_cond(sbi, NULL, NULL, ino, NODE);
return ret ? -EIO: 0;
}
f2fs: fix to writeout dirty inode during node flush As Eric reported: On xfstest generic/204 on f2fs, I'm getting a kernel BUG. allocate_segment_by_default+0x9d/0x100 [f2fs] f2fs_allocate_data_block+0x3c0/0x5c0 [f2fs] do_write_page+0x62/0x110 [f2fs] f2fs_do_write_node_page+0x2b/0xa0 [f2fs] __write_node_page+0x2ec/0x590 [f2fs] f2fs_sync_node_pages+0x756/0x7e0 [f2fs] block_operations+0x25b/0x350 [f2fs] f2fs_write_checkpoint+0x104/0x1150 [f2fs] f2fs_sync_fs+0xa2/0x120 [f2fs] f2fs_balance_fs_bg+0x33c/0x390 [f2fs] f2fs_write_node_pages+0x4c/0x1f0 [f2fs] do_writepages+0x1c/0x70 __writeback_single_inode+0x45/0x320 writeback_sb_inodes+0x273/0x5c0 wb_writeback+0xff/0x2e0 wb_workfn+0xa1/0x370 process_one_work+0x138/0x350 worker_thread+0x4d/0x3d0 kthread+0x109/0x140 The root cause of this issue is, in a very small partition, e.g. in generic/204 testcase of fstest suit, filesystem's free space is 50MB, so at most we can write 12800 inline inode with command: `echo XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX > $SCRATCH_MNT/$i`, then filesystem will have: - 12800 dirty inline data page - 12800 dirty inode page - and 12800 dirty imeta (dirty inode) When we flush node-inode's page cache, we can also flush inline data with each inode page, however it will run out-of-free-space in device, then once it triggers checkpoint, there is no room for huge number of imeta, at this time, GC is useless, as there is no dirty segment at all. In order to fix this, we try to recognize inode page during node_inode's page flushing, and update inode page from dirty inode, so that later another imeta (dirty inode) flush can be avoided. Reported-and-tested-by: Eric Biggers <ebiggers@kernel.org> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-08-22 12:17:56 +00:00
static int f2fs_match_ino(struct inode *inode, unsigned long ino, void *data)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
bool clean;
if (inode->i_ino != ino)
return 0;
if (!is_inode_flag_set(inode, FI_DIRTY_INODE))
return 0;
spin_lock(&sbi->inode_lock[DIRTY_META]);
clean = list_empty(&F2FS_I(inode)->gdirty_list);
spin_unlock(&sbi->inode_lock[DIRTY_META]);
if (clean)
return 0;
inode = igrab(inode);
if (!inode)
return 0;
return 1;
}
static bool flush_dirty_inode(struct page *page)
{
struct f2fs_sb_info *sbi = F2FS_P_SB(page);
struct inode *inode;
nid_t ino = ino_of_node(page);
inode = find_inode_nowait(sbi->sb, ino, f2fs_match_ino, NULL);
if (!inode)
return false;
f2fs_update_inode(inode, page);
unlock_page(page);
iput(inode);
return true;
}
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
struct writeback_control *wbc,
bool do_balance, enum iostat_type io_type)
{
pgoff_t index;
struct pagevec pvec;
int step = 0;
int nwritten = 0;
int ret = 0;
int nr_pages, done = 0;
pagevec_init(&pvec);
next_step:
index = 0;
while (!done && (nr_pages = pagevec_lookup_tag(&pvec,
NODE_MAPPING(sbi), &index, PAGECACHE_TAG_DIRTY))) {
int i;
for (i = 0; i < nr_pages; i++) {
struct page *page = pvec.pages[i];
bool submitted = false;
f2fs: fix to writeout dirty inode during node flush As Eric reported: On xfstest generic/204 on f2fs, I'm getting a kernel BUG. allocate_segment_by_default+0x9d/0x100 [f2fs] f2fs_allocate_data_block+0x3c0/0x5c0 [f2fs] do_write_page+0x62/0x110 [f2fs] f2fs_do_write_node_page+0x2b/0xa0 [f2fs] __write_node_page+0x2ec/0x590 [f2fs] f2fs_sync_node_pages+0x756/0x7e0 [f2fs] block_operations+0x25b/0x350 [f2fs] f2fs_write_checkpoint+0x104/0x1150 [f2fs] f2fs_sync_fs+0xa2/0x120 [f2fs] f2fs_balance_fs_bg+0x33c/0x390 [f2fs] f2fs_write_node_pages+0x4c/0x1f0 [f2fs] do_writepages+0x1c/0x70 __writeback_single_inode+0x45/0x320 writeback_sb_inodes+0x273/0x5c0 wb_writeback+0xff/0x2e0 wb_workfn+0xa1/0x370 process_one_work+0x138/0x350 worker_thread+0x4d/0x3d0 kthread+0x109/0x140 The root cause of this issue is, in a very small partition, e.g. in generic/204 testcase of fstest suit, filesystem's free space is 50MB, so at most we can write 12800 inline inode with command: `echo XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX > $SCRATCH_MNT/$i`, then filesystem will have: - 12800 dirty inline data page - 12800 dirty inode page - and 12800 dirty imeta (dirty inode) When we flush node-inode's page cache, we can also flush inline data with each inode page, however it will run out-of-free-space in device, then once it triggers checkpoint, there is no room for huge number of imeta, at this time, GC is useless, as there is no dirty segment at all. In order to fix this, we try to recognize inode page during node_inode's page flushing, and update inode page from dirty inode, so that later another imeta (dirty inode) flush can be avoided. Reported-and-tested-by: Eric Biggers <ebiggers@kernel.org> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-08-22 12:17:56 +00:00
bool may_dirty = true;
/* give a priority to WB_SYNC threads */
if (atomic_read(&sbi->wb_sync_req[NODE]) &&
wbc->sync_mode == WB_SYNC_NONE) {
done = 1;
break;
}
/*
* flushing sequence with step:
* 0. indirect nodes
* 1. dentry dnodes
* 2. file dnodes
*/
if (step == 0 && IS_DNODE(page))
continue;
if (step == 1 && (!IS_DNODE(page) ||
is_cold_node(page)))
continue;
if (step == 2 && (!IS_DNODE(page) ||
!is_cold_node(page)))
continue;
lock_node:
if (wbc->sync_mode == WB_SYNC_ALL)
lock_page(page);
else if (!trylock_page(page))
continue;
if (unlikely(page->mapping != NODE_MAPPING(sbi))) {
continue_unlock:
unlock_page(page);
continue;
}
if (!PageDirty(page)) {
/* someone wrote it for us */
goto continue_unlock;
}
/* flush inline_data */
if (is_inline_node(page)) {
clear_inline_node(page);
unlock_page(page);
flush_inline_data(sbi, ino_of_node(page));
goto lock_node;
f2fs: fix to writeout dirty inode during node flush As Eric reported: On xfstest generic/204 on f2fs, I'm getting a kernel BUG. allocate_segment_by_default+0x9d/0x100 [f2fs] f2fs_allocate_data_block+0x3c0/0x5c0 [f2fs] do_write_page+0x62/0x110 [f2fs] f2fs_do_write_node_page+0x2b/0xa0 [f2fs] __write_node_page+0x2ec/0x590 [f2fs] f2fs_sync_node_pages+0x756/0x7e0 [f2fs] block_operations+0x25b/0x350 [f2fs] f2fs_write_checkpoint+0x104/0x1150 [f2fs] f2fs_sync_fs+0xa2/0x120 [f2fs] f2fs_balance_fs_bg+0x33c/0x390 [f2fs] f2fs_write_node_pages+0x4c/0x1f0 [f2fs] do_writepages+0x1c/0x70 __writeback_single_inode+0x45/0x320 writeback_sb_inodes+0x273/0x5c0 wb_writeback+0xff/0x2e0 wb_workfn+0xa1/0x370 process_one_work+0x138/0x350 worker_thread+0x4d/0x3d0 kthread+0x109/0x140 The root cause of this issue is, in a very small partition, e.g. in generic/204 testcase of fstest suit, filesystem's free space is 50MB, so at most we can write 12800 inline inode with command: `echo XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX > $SCRATCH_MNT/$i`, then filesystem will have: - 12800 dirty inline data page - 12800 dirty inode page - and 12800 dirty imeta (dirty inode) When we flush node-inode's page cache, we can also flush inline data with each inode page, however it will run out-of-free-space in device, then once it triggers checkpoint, there is no room for huge number of imeta, at this time, GC is useless, as there is no dirty segment at all. In order to fix this, we try to recognize inode page during node_inode's page flushing, and update inode page from dirty inode, so that later another imeta (dirty inode) flush can be avoided. Reported-and-tested-by: Eric Biggers <ebiggers@kernel.org> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-08-22 12:17:56 +00:00
}
/* flush dirty inode */
if (IS_INODE(page) && may_dirty) {
may_dirty = false;
if (flush_dirty_inode(page))
goto lock_node;
}
f2fs_wait_on_page_writeback(page, NODE, true, true);
if (!clear_page_dirty_for_io(page))
goto continue_unlock;
set_fsync_mark(page, 0);
set_dentry_mark(page, 0);
ret = __write_node_page(page, false, &submitted,
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
wbc, do_balance, io_type, NULL);
if (ret)
unlock_page(page);
else if (submitted)
nwritten++;
if (--wbc->nr_to_write == 0)
break;
}
pagevec_release(&pvec);
cond_resched();
if (wbc->nr_to_write == 0) {
step = 2;
break;
}
}
if (step < 2) {
if (!is_sbi_flag_set(sbi, SBI_CP_DISABLED) &&
wbc->sync_mode == WB_SYNC_NONE && step == 1)
goto out;
step++;
goto next_step;
}
out:
if (nwritten)
f2fs_submit_merged_write(sbi, NODE);
if (unlikely(f2fs_cp_error(sbi)))
return -EIO;
return ret;
}
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
unsigned int seq_id)
{
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
struct fsync_node_entry *fn;
struct page *page;
struct list_head *head = &sbi->fsync_node_list;
unsigned long flags;
unsigned int cur_seq_id = 0;
int ret2, ret = 0;
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
while (seq_id && cur_seq_id < seq_id) {
spin_lock_irqsave(&sbi->fsync_node_lock, flags);
if (list_empty(head)) {
spin_unlock_irqrestore(&sbi->fsync_node_lock, flags);
break;
}
fn = list_first_entry(head, struct fsync_node_entry, list);
if (fn->seq_id > seq_id) {
spin_unlock_irqrestore(&sbi->fsync_node_lock, flags);
break;
}
cur_seq_id = fn->seq_id;
page = fn->page;
get_page(page);
spin_unlock_irqrestore(&sbi->fsync_node_lock, flags);
f2fs_wait_on_page_writeback(page, NODE, true, false);
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
if (TestClearPageError(page))
ret = -EIO;
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
put_page(page);
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
if (ret)
break;
}
ret2 = filemap_check_errors(NODE_MAPPING(sbi));
if (!ret)
ret = ret2;
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
return ret;
}
static int f2fs_write_node_pages(struct address_space *mapping,
struct writeback_control *wbc)
{
struct f2fs_sb_info *sbi = F2FS_M_SB(mapping);
struct blk_plug plug;
long diff;
if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
goto skip_write;
/* balancing f2fs's metadata in background */
f2fs_balance_fs_bg(sbi);
/* collect a number of dirty node pages and write together */
if (wbc->sync_mode != WB_SYNC_ALL &&
get_pages(sbi, F2FS_DIRTY_NODES) <
nr_pages_to_skip(sbi, NODE))
goto skip_write;
if (wbc->sync_mode == WB_SYNC_ALL)
atomic_inc(&sbi->wb_sync_req[NODE]);
else if (atomic_read(&sbi->wb_sync_req[NODE]))
goto skip_write;
trace_f2fs_writepages(mapping->host, wbc, NODE);
diff = nr_pages_to_write(sbi, NODE, wbc);
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_node_pages(sbi, wbc, true, FS_NODE_IO);
blk_finish_plug(&plug);
wbc->nr_to_write = max((long)0, wbc->nr_to_write - diff);
if (wbc->sync_mode == WB_SYNC_ALL)
atomic_dec(&sbi->wb_sync_req[NODE]);
return 0;
skip_write:
wbc->pages_skipped += get_pages(sbi, F2FS_DIRTY_NODES);
trace_f2fs_writepages(mapping->host, wbc, NODE);
return 0;
}
static int f2fs_set_node_page_dirty(struct page *page)
{
trace_f2fs_set_page_dirty(page, NODE);
if (!PageUptodate(page))
SetPageUptodate(page);
#ifdef CONFIG_F2FS_CHECK_FS
if (IS_INODE(page))
f2fs_inode_chksum_set(F2FS_P_SB(page), page);
#endif
if (!PageDirty(page)) {
__set_page_dirty_nobuffers(page);
inc_page_count(F2FS_P_SB(page), F2FS_DIRTY_NODES);
f2fs_set_page_private(page, 0);
f2fs_trace_pid(page);
return 1;
}
return 0;
}
/*
* Structure of the f2fs node operations
*/
const struct address_space_operations f2fs_node_aops = {
.writepage = f2fs_write_node_page,
.writepages = f2fs_write_node_pages,
.set_page_dirty = f2fs_set_node_page_dirty,
.invalidatepage = f2fs_invalidate_page,
.releasepage = f2fs_release_page,
#ifdef CONFIG_MIGRATION
.migratepage = f2fs_migrate_page,
#endif
};
static struct free_nid *__lookup_free_nid_list(struct f2fs_nm_info *nm_i,
nid_t n)
{
return radix_tree_lookup(&nm_i->free_nid_root, n);
}
static int __insert_free_nid(struct f2fs_sb_info *sbi,
struct free_nid *i, enum nid_state state)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
int err = radix_tree_insert(&nm_i->free_nid_root, i->nid, i);
if (err)
return err;
f2fs_bug_on(sbi, state != i->state);
nm_i->nid_cnt[state]++;
if (state == FREE_NID)
list_add_tail(&i->list, &nm_i->free_nid_list);
return 0;
}
static void __remove_free_nid(struct f2fs_sb_info *sbi,
struct free_nid *i, enum nid_state state)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
f2fs_bug_on(sbi, state != i->state);
nm_i->nid_cnt[state]--;
if (state == FREE_NID)
list_del(&i->list);
radix_tree_delete(&nm_i->free_nid_root, i->nid);
}
static void __move_free_nid(struct f2fs_sb_info *sbi, struct free_nid *i,
enum nid_state org_state, enum nid_state dst_state)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
f2fs_bug_on(sbi, org_state != i->state);
i->state = dst_state;
nm_i->nid_cnt[org_state]--;
nm_i->nid_cnt[dst_state]++;
switch (dst_state) {
case PREALLOC_NID:
list_del(&i->list);
break;
case FREE_NID:
list_add_tail(&i->list, &nm_i->free_nid_list);
break;
default:
BUG_ON(1);
}
}
static void update_free_nid_bitmap(struct f2fs_sb_info *sbi, nid_t nid,
bool set, bool build)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
unsigned int nat_ofs = NAT_BLOCK_OFFSET(nid);
unsigned int nid_ofs = nid - START_NID(nid);
if (!test_bit_le(nat_ofs, nm_i->nat_block_bitmap))
return;
if (set) {
if (test_bit_le(nid_ofs, nm_i->free_nid_bitmap[nat_ofs]))
return;
__set_bit_le(nid_ofs, nm_i->free_nid_bitmap[nat_ofs]);
nm_i->free_nid_count[nat_ofs]++;
} else {
if (!test_bit_le(nid_ofs, nm_i->free_nid_bitmap[nat_ofs]))
return;
__clear_bit_le(nid_ofs, nm_i->free_nid_bitmap[nat_ofs]);
if (!build)
nm_i->free_nid_count[nat_ofs]--;
}
}
/* return if the nid is recognized as free */
static bool add_free_nid(struct f2fs_sb_info *sbi,
nid_t nid, bool build, bool update)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct free_nid *i, *e;
struct nat_entry *ne;
int err = -EINVAL;
bool ret = false;
/* 0 nid should not be used */
if (unlikely(nid == 0))
return false;
if (unlikely(f2fs_check_nid_range(sbi, nid)))
return false;
i = f2fs_kmem_cache_alloc(free_nid_slab, GFP_NOFS);
i->nid = nid;
i->state = FREE_NID;
radix_tree_preload(GFP_NOFS | __GFP_NOFAIL);
spin_lock(&nm_i->nid_list_lock);
if (build) {
/*
* Thread A Thread B
* - f2fs_create
* - f2fs_new_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_alloc_nid
* - __insert_nid_to_list(PREALLOC_NID)
* - f2fs_balance_fs_bg
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
* - __f2fs_build_free_nids
* - scan_nat_page
* - add_free_nid
* - __lookup_nat_cache
* - f2fs_add_link
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_init_inode_metadata
* - f2fs_new_inode_page
* - f2fs_new_node_page
* - set_node_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_alloc_nid_done
* - __remove_nid_from_list(PREALLOC_NID)
* - __insert_nid_to_list(FREE_NID)
*/
ne = __lookup_nat_cache(nm_i, nid);
if (ne && (!get_nat_flag(ne, IS_CHECKPOINTED) ||
nat_get_blkaddr(ne) != NULL_ADDR))
goto err_out;
e = __lookup_free_nid_list(nm_i, nid);
if (e) {
if (e->state == FREE_NID)
ret = true;
goto err_out;
}
}
ret = true;
err = __insert_free_nid(sbi, i, FREE_NID);
err_out:
if (update) {
update_free_nid_bitmap(sbi, nid, ret, build);
if (!build)
nm_i->available_nids++;
}
spin_unlock(&nm_i->nid_list_lock);
radix_tree_preload_end();
if (err)
kmem_cache_free(free_nid_slab, i);
return ret;
}
static void remove_free_nid(struct f2fs_sb_info *sbi, nid_t nid)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct free_nid *i;
bool need_free = false;
spin_lock(&nm_i->nid_list_lock);
i = __lookup_free_nid_list(nm_i, nid);
if (i && i->state == FREE_NID) {
__remove_free_nid(sbi, i, FREE_NID);
need_free = true;
}
spin_unlock(&nm_i->nid_list_lock);
if (need_free)
kmem_cache_free(free_nid_slab, i);
}
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
static int scan_nat_page(struct f2fs_sb_info *sbi,
struct page *nat_page, nid_t start_nid)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct f2fs_nat_block *nat_blk = page_address(nat_page);
block_t blk_addr;
unsigned int nat_ofs = NAT_BLOCK_OFFSET(start_nid);
int i;
__set_bit_le(nat_ofs, nm_i->nat_block_bitmap);
i = start_nid % NAT_ENTRY_PER_BLOCK;
for (; i < NAT_ENTRY_PER_BLOCK; i++, start_nid++) {
if (unlikely(start_nid >= nm_i->max_nid))
break;
blk_addr = le32_to_cpu(nat_blk->entries[i].block_addr);
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
if (blk_addr == NEW_ADDR)
return -EINVAL;
if (blk_addr == NULL_ADDR) {
add_free_nid(sbi, start_nid, true, true);
} else {
spin_lock(&NM_I(sbi)->nid_list_lock);
update_free_nid_bitmap(sbi, start_nid, false, true);
spin_unlock(&NM_I(sbi)->nid_list_lock);
}
}
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
return 0;
}
static void scan_curseg_cache(struct f2fs_sb_info *sbi)
{
struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA);
struct f2fs_journal *journal = curseg->journal;
int i;
down_read(&curseg->journal_rwsem);
for (i = 0; i < nats_in_cursum(journal); i++) {
block_t addr;
nid_t nid;
addr = le32_to_cpu(nat_in_journal(journal, i).block_addr);
nid = le32_to_cpu(nid_in_journal(journal, i));
if (addr == NULL_ADDR)
add_free_nid(sbi, nid, true, false);
else
remove_free_nid(sbi, nid);
}
up_read(&curseg->journal_rwsem);
}
static void scan_free_nid_bits(struct f2fs_sb_info *sbi)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
unsigned int i, idx;
nid_t nid;
down_read(&nm_i->nat_tree_lock);
for (i = 0; i < nm_i->nat_blocks; i++) {
if (!test_bit_le(i, nm_i->nat_block_bitmap))
continue;
if (!nm_i->free_nid_count[i])
continue;
for (idx = 0; idx < NAT_ENTRY_PER_BLOCK; idx++) {
idx = find_next_bit_le(nm_i->free_nid_bitmap[i],
NAT_ENTRY_PER_BLOCK, idx);
if (idx >= NAT_ENTRY_PER_BLOCK)
break;
nid = i * NAT_ENTRY_PER_BLOCK + idx;
add_free_nid(sbi, nid, true, false);
if (nm_i->nid_cnt[FREE_NID] >= MAX_FREE_NIDS)
goto out;
}
}
out:
scan_curseg_cache(sbi);
up_read(&nm_i->nat_tree_lock);
}
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
static int __f2fs_build_free_nids(struct f2fs_sb_info *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
bool sync, bool mount)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
int i = 0, ret;
nid_t nid = nm_i->next_scan_nid;
if (unlikely(nid >= nm_i->max_nid))
nid = 0;
/* Enough entries */
if (nm_i->nid_cnt[FREE_NID] >= NAT_ENTRY_PER_BLOCK)
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
return 0;
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
if (!sync && !f2fs_available_free_memory(sbi, FREE_NIDS))
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
return 0;
if (!mount) {
/* try to find free nids in free_nid_bitmap */
scan_free_nid_bits(sbi);
if (nm_i->nid_cnt[FREE_NID] >= NAT_ENTRY_PER_BLOCK)
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
return 0;
}
/* readahead nat pages to be scanned */
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, NAT_BLOCK_OFFSET(nid), FREE_NID_PAGES,
META_NAT, true);
down_read(&nm_i->nat_tree_lock);
while (1) {
if (!test_bit_le(NAT_BLOCK_OFFSET(nid),
nm_i->nat_block_bitmap)) {
struct page *page = get_current_nat_page(sbi, nid);
if (IS_ERR(page)) {
ret = PTR_ERR(page);
} else {
ret = scan_nat_page(sbi, page, nid);
f2fs_put_page(page, 1);
}
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
if (ret) {
up_read(&nm_i->nat_tree_lock);
f2fs_err(sbi, "NAT is corrupt, run fsck to fix it");
return ret;
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
}
}
nid += (NAT_ENTRY_PER_BLOCK - (nid % NAT_ENTRY_PER_BLOCK));
if (unlikely(nid >= nm_i->max_nid))
nid = 0;
if (++i >= FREE_NID_PAGES)
break;
}
/* go to the next free nat pages to find free nids abundantly */
nm_i->next_scan_nid = nid;
/* find free nids from current sum_pages */
scan_curseg_cache(sbi);
up_read(&nm_i->nat_tree_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
f2fs_ra_meta_pages(sbi, NAT_BLOCK_OFFSET(nm_i->next_scan_nid),
nm_i->ra_nid_pages, META_NAT, false);
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
return 0;
}
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount)
{
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
int ret;
mutex_lock(&NM_I(sbi)->build_lock);
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
ret = __f2fs_build_free_nids(sbi, sync, mount);
mutex_unlock(&NM_I(sbi)->build_lock);
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
return ret;
}
/*
* If this function returns success, caller can obtain a new nid
* from second parameter of this function.
* The returned nid could be used ino as well as nid when inode is created.
*/
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_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct free_nid *i = NULL;
retry:
if (time_to_inject(sbi, FAULT_ALLOC_NID)) {
f2fs_show_injection_info(sbi, FAULT_ALLOC_NID);
return false;
}
spin_lock(&nm_i->nid_list_lock);
if (unlikely(nm_i->available_nids == 0)) {
spin_unlock(&nm_i->nid_list_lock);
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
/* We should not use stale free nids created by f2fs_build_free_nids */
if (nm_i->nid_cnt[FREE_NID] && !on_f2fs_build_free_nids(nm_i)) {
f2fs_bug_on(sbi, list_empty(&nm_i->free_nid_list));
i = list_first_entry(&nm_i->free_nid_list,
struct free_nid, list);
*nid = i->nid;
__move_free_nid(sbi, i, FREE_NID, PREALLOC_NID);
nm_i->available_nids--;
update_free_nid_bitmap(sbi, *nid, false, false);
spin_unlock(&nm_i->nid_list_lock);
return true;
}
spin_unlock(&nm_i->nid_list_lock);
/* Let's scan nat pages and its caches to get free nids */
if (!f2fs_build_free_nids(sbi, true, false))
goto retry;
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
* f2fs_alloc_nid() should be called prior to this function.
*/
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_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct free_nid *i;
spin_lock(&nm_i->nid_list_lock);
i = __lookup_free_nid_list(nm_i, nid);
f2fs_bug_on(sbi, !i);
__remove_free_nid(sbi, i, PREALLOC_NID);
spin_unlock(&nm_i->nid_list_lock);
kmem_cache_free(free_nid_slab, 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
* f2fs_alloc_nid() should be called prior to this function.
*/
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_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct free_nid *i;
bool need_free = false;
if (!nid)
return;
spin_lock(&nm_i->nid_list_lock);
i = __lookup_free_nid_list(nm_i, nid);
f2fs_bug_on(sbi, !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_available_free_memory(sbi, FREE_NIDS)) {
__remove_free_nid(sbi, i, PREALLOC_NID);
need_free = true;
} else {
__move_free_nid(sbi, i, PREALLOC_NID, FREE_NID);
}
nm_i->available_nids++;
update_free_nid_bitmap(sbi, nid, true, false);
spin_unlock(&nm_i->nid_list_lock);
if (need_free)
kmem_cache_free(free_nid_slab, 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
int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct free_nid *i, *next;
int nr = nr_shrink;
if (nm_i->nid_cnt[FREE_NID] <= MAX_FREE_NIDS)
return 0;
if (!mutex_trylock(&nm_i->build_lock))
return 0;
spin_lock(&nm_i->nid_list_lock);
list_for_each_entry_safe(i, next, &nm_i->free_nid_list, list) {
if (nr_shrink <= 0 ||
nm_i->nid_cnt[FREE_NID] <= MAX_FREE_NIDS)
break;
__remove_free_nid(sbi, i, FREE_NID);
kmem_cache_free(free_nid_slab, i);
nr_shrink--;
}
spin_unlock(&nm_i->nid_list_lock);
mutex_unlock(&nm_i->build_lock);
return nr - nr_shrink;
}
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_recover_inline_xattr(struct inode *inode, struct page *page)
{
void *src_addr, *dst_addr;
size_t inline_size;
struct page *ipage;
struct f2fs_inode *ri;
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
ipage = f2fs_get_node_page(F2FS_I_SB(inode), inode->i_ino);
f2fs_bug_on(F2FS_I_SB(inode), IS_ERR(ipage));
ri = F2FS_INODE(page);
if (ri->i_inline & F2FS_INLINE_XATTR) {
set_inode_flag(inode, FI_INLINE_XATTR);
} else {
clear_inode_flag(inode, FI_INLINE_XATTR);
goto update_inode;
}
f2fs: support flexible inline xattr size Now, in product, more and more features based on file encryption were introduced, their demand of xattr space is increasing, however, inline xattr has fixed-size of 200 bytes, once inline xattr space is full, new increased xattr data would occupy additional xattr block which may bring us more space usage and performance regression during persisting. In order to resolve above issue, it's better to expand inline xattr size flexibly according to user's requirement. So this patch introduces new filesystem feature 'flexible inline xattr', and new mount option 'inline_xattr_size=%u', once mkfs enables the feature, we can use the option to make f2fs supporting flexible inline xattr size. To support this feature, we add extra attribute i_inline_xattr_size in inode layout, indicating that how many space inline xattr borrows from block address mapping space in inode layout, by this, we can easily locate and store flexible-sized inline xattr data in inode. Inode disk layout: +----------------------+ | .i_mode | | ... | | .i_ext | +----------------------+ | .i_extra_isize | | .i_inline_xattr_size |-----------+ | ... | | +----------------------+ | | .i_addr | | | - block address or | | | - inline data | | +----------------------+<---+ v | inline xattr | +---inline xattr range +----------------------+<---+ | .i_nid | +----------------------+ | node_footer | | (nid, ino, offset) | +----------------------+ Note that, we have to cnosider backward compatibility which reserved inline_data space, 200 bytes, all the time, reported by Sheng Yong. Previous inline data or directory always reserved 200 bytes in inode layout, even if inline_xattr is disabled. In order to keep inline_dentry's structure for backward compatibility, we get the space back only from inline_data. Signed-off-by: Chao Yu <yuchao0@huawei.com> Reported-by: Sheng Yong <shengyong1@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-09-06 13:59:50 +00:00
dst_addr = inline_xattr_addr(inode, ipage);
src_addr = inline_xattr_addr(inode, page);
inline_size = inline_xattr_size(inode);
f2fs_wait_on_page_writeback(ipage, NODE, true, true);
memcpy(dst_addr, src_addr, inline_size);
update_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_update_inode(inode, ipage);
f2fs_put_page(ipage, 1);
}
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_recover_xattr_data(struct inode *inode, struct page *page)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
nid_t prev_xnid = F2FS_I(inode)->i_xattr_nid;
nid_t new_xnid;
struct dnode_of_data dn;
struct node_info ni;
struct page *xpage;
int err;
if (!prev_xnid)
goto recover_xnid;
/* 1: invalidate the previous xattr nid */
err = f2fs_get_node_info(sbi, prev_xnid, &ni);
if (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
f2fs_invalidate_blocks(sbi, ni.blk_addr);
dec_valid_node_count(sbi, inode, false);
set_node_addr(sbi, &ni, NULL_ADDR, false);
recover_xnid:
/* 2: update xattr nid in 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
if (!f2fs_alloc_nid(sbi, &new_xnid))
return -ENOSPC;
set_new_dnode(&dn, inode, NULL, NULL, new_xnid);
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
xpage = f2fs_new_node_page(&dn, XATTR_NODE_OFFSET);
if (IS_ERR(xpage)) {
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_alloc_nid_failed(sbi, new_xnid);
return PTR_ERR(xpage);
}
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_alloc_nid_done(sbi, new_xnid);
f2fs_update_inode_page(inode);
/* 3: update and set xattr node page dirty */
memcpy(F2FS_NODE(xpage), F2FS_NODE(page), VALID_XATTR_BLOCK_SIZE);
set_page_dirty(xpage);
f2fs_put_page(xpage, 1);
return 0;
}
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page)
{
struct f2fs_inode *src, *dst;
nid_t ino = ino_of_node(page);
struct node_info old_ni, new_ni;
struct page *ipage;
int err;
err = f2fs_get_node_info(sbi, ino, &old_ni);
if (err)
return err;
if (unlikely(old_ni.blk_addr != NULL_ADDR))
return -EINVAL;
retry:
ipage = f2fs_grab_cache_page(NODE_MAPPING(sbi), ino, false);
if (!ipage) {
congestion_wait(BLK_RW_ASYNC, HZ/50);
goto retry;
}
/* Should not use this inode from free nid list */
remove_free_nid(sbi, ino);
if (!PageUptodate(ipage))
SetPageUptodate(ipage);
fill_node_footer(ipage, ino, ino, 0, true);
set_cold_node(ipage, false);
src = F2FS_INODE(page);
dst = F2FS_INODE(ipage);
memcpy(dst, src, (unsigned long)&src->i_ext - (unsigned long)src);
dst->i_size = 0;
dst->i_blocks = cpu_to_le64(1);
dst->i_links = cpu_to_le32(1);
dst->i_xattr_nid = 0;
f2fs: enhance on-disk inode structure scalability This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline to indicate that on-disk structure of current inode is extended. In order to extend, we changed the inode structure a bit: Original one: struct f2fs_inode { ... struct f2fs_extent i_ext; __le32 i_addr[DEF_ADDRS_PER_INODE]; __le32 i_nid[DEF_NIDS_PER_INODE]; } Extended one: struct f2fs_inode { ... struct f2fs_extent i_ext; union { struct { __le16 i_extra_isize; __le16 i_padding; __le32 i_extra_end[0]; }; __le32 i_addr[DEF_ADDRS_PER_INODE]; }; __le32 i_nid[DEF_NIDS_PER_INODE]; } Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of i_addr field for storing i_extra_isize and i_padding. with i_extra_isize, we can calculate actual size of reserved space in i_addr, available attribute fields included in total extra attribute fields for current inode can be described as below: +--------------------+ | .i_mode | | ... | | .i_ext | +--------------------+ | .i_extra_isize |-----+ | .i_padding | | | .i_prjid | | | .i_atime_extra | | | .i_ctime_extra | | | .i_mtime_extra |<----+ | .i_inode_cs |<----- store blkaddr/inline from here | .i_xattr_cs | | ... | +--------------------+ | | | block address | | | +--------------------+ | .i_nid | +--------------------+ | node_footer | | (nid, ino, offset) | +--------------------+ Hence, with this patch, we would enhance scalability of f2fs inode for storing more newly added attribute. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-18 16:19:06 +00:00
dst->i_inline = src->i_inline & (F2FS_INLINE_XATTR | F2FS_EXTRA_ATTR);
if (dst->i_inline & F2FS_EXTRA_ATTR) {
f2fs: enhance on-disk inode structure scalability This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline to indicate that on-disk structure of current inode is extended. In order to extend, we changed the inode structure a bit: Original one: struct f2fs_inode { ... struct f2fs_extent i_ext; __le32 i_addr[DEF_ADDRS_PER_INODE]; __le32 i_nid[DEF_NIDS_PER_INODE]; } Extended one: struct f2fs_inode { ... struct f2fs_extent i_ext; union { struct { __le16 i_extra_isize; __le16 i_padding; __le32 i_extra_end[0]; }; __le32 i_addr[DEF_ADDRS_PER_INODE]; }; __le32 i_nid[DEF_NIDS_PER_INODE]; } Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of i_addr field for storing i_extra_isize and i_padding. with i_extra_isize, we can calculate actual size of reserved space in i_addr, available attribute fields included in total extra attribute fields for current inode can be described as below: +--------------------+ | .i_mode | | ... | | .i_ext | +--------------------+ | .i_extra_isize |-----+ | .i_padding | | | .i_prjid | | | .i_atime_extra | | | .i_ctime_extra | | | .i_mtime_extra |<----+ | .i_inode_cs |<----- store blkaddr/inline from here | .i_xattr_cs | | ... | +--------------------+ | | | block address | | | +--------------------+ | .i_nid | +--------------------+ | node_footer | | (nid, ino, offset) | +--------------------+ Hence, with this patch, we would enhance scalability of f2fs inode for storing more newly added attribute. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-18 16:19:06 +00:00
dst->i_extra_isize = src->i_extra_isize;
f2fs: support flexible inline xattr size Now, in product, more and more features based on file encryption were introduced, their demand of xattr space is increasing, however, inline xattr has fixed-size of 200 bytes, once inline xattr space is full, new increased xattr data would occupy additional xattr block which may bring us more space usage and performance regression during persisting. In order to resolve above issue, it's better to expand inline xattr size flexibly according to user's requirement. So this patch introduces new filesystem feature 'flexible inline xattr', and new mount option 'inline_xattr_size=%u', once mkfs enables the feature, we can use the option to make f2fs supporting flexible inline xattr size. To support this feature, we add extra attribute i_inline_xattr_size in inode layout, indicating that how many space inline xattr borrows from block address mapping space in inode layout, by this, we can easily locate and store flexible-sized inline xattr data in inode. Inode disk layout: +----------------------+ | .i_mode | | ... | | .i_ext | +----------------------+ | .i_extra_isize | | .i_inline_xattr_size |-----------+ | ... | | +----------------------+ | | .i_addr | | | - block address or | | | - inline data | | +----------------------+<---+ v | inline xattr | +---inline xattr range +----------------------+<---+ | .i_nid | +----------------------+ | node_footer | | (nid, ino, offset) | +----------------------+ Note that, we have to cnosider backward compatibility which reserved inline_data space, 200 bytes, all the time, reported by Sheng Yong. Previous inline data or directory always reserved 200 bytes in inode layout, even if inline_xattr is disabled. In order to keep inline_dentry's structure for backward compatibility, we get the space back only from inline_data. Signed-off-by: Chao Yu <yuchao0@huawei.com> Reported-by: Sheng Yong <shengyong1@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-09-06 13:59:50 +00:00
if (f2fs_sb_has_flexible_inline_xattr(sbi) &&
f2fs: support flexible inline xattr size Now, in product, more and more features based on file encryption were introduced, their demand of xattr space is increasing, however, inline xattr has fixed-size of 200 bytes, once inline xattr space is full, new increased xattr data would occupy additional xattr block which may bring us more space usage and performance regression during persisting. In order to resolve above issue, it's better to expand inline xattr size flexibly according to user's requirement. So this patch introduces new filesystem feature 'flexible inline xattr', and new mount option 'inline_xattr_size=%u', once mkfs enables the feature, we can use the option to make f2fs supporting flexible inline xattr size. To support this feature, we add extra attribute i_inline_xattr_size in inode layout, indicating that how many space inline xattr borrows from block address mapping space in inode layout, by this, we can easily locate and store flexible-sized inline xattr data in inode. Inode disk layout: +----------------------+ | .i_mode | | ... | | .i_ext | +----------------------+ | .i_extra_isize | | .i_inline_xattr_size |-----------+ | ... | | +----------------------+ | | .i_addr | | | - block address or | | | - inline data | | +----------------------+<---+ v | inline xattr | +---inline xattr range +----------------------+<---+ | .i_nid | +----------------------+ | node_footer | | (nid, ino, offset) | +----------------------+ Note that, we have to cnosider backward compatibility which reserved inline_data space, 200 bytes, all the time, reported by Sheng Yong. Previous inline data or directory always reserved 200 bytes in inode layout, even if inline_xattr is disabled. In order to keep inline_dentry's structure for backward compatibility, we get the space back only from inline_data. Signed-off-by: Chao Yu <yuchao0@huawei.com> Reported-by: Sheng Yong <shengyong1@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-09-06 13:59:50 +00:00
F2FS_FITS_IN_INODE(src, le16_to_cpu(src->i_extra_isize),
i_inline_xattr_size))
dst->i_inline_xattr_size = src->i_inline_xattr_size;
if (f2fs_sb_has_project_quota(sbi) &&
F2FS_FITS_IN_INODE(src, le16_to_cpu(src->i_extra_isize),
i_projid))
dst->i_projid = src->i_projid;
if (f2fs_sb_has_inode_crtime(sbi) &&
F2FS_FITS_IN_INODE(src, le16_to_cpu(src->i_extra_isize),
i_crtime_nsec)) {
dst->i_crtime = src->i_crtime;
dst->i_crtime_nsec = src->i_crtime_nsec;
}
}
new_ni = old_ni;
new_ni.ino = ino;
if (unlikely(inc_valid_node_count(sbi, NULL, true)))
WARN_ON(1);
set_node_addr(sbi, &new_ni, NEW_ADDR, false);
inc_valid_inode_count(sbi);
set_page_dirty(ipage);
f2fs_put_page(ipage, 1);
return 0;
}
int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
unsigned int segno, struct f2fs_summary_block *sum)
{
struct f2fs_node *rn;
struct f2fs_summary *sum_entry;
block_t addr;
int i, idx, last_offset, nrpages;
/* scan the node segment */
last_offset = sbi->blocks_per_seg;
addr = START_BLOCK(sbi, segno);
sum_entry = &sum->entries[0];
for (i = 0; i < last_offset; i += nrpages, addr += nrpages) {
nrpages = min(last_offset - i, BIO_MAX_PAGES);
/* readahead node pages */
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, addr, nrpages, META_POR, true);
for (idx = addr; idx < addr + nrpages; idx++) {
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_get_tmp_page(sbi, idx);
if (IS_ERR(page))
return PTR_ERR(page);
rn = F2FS_NODE(page);
sum_entry->nid = rn->footer.nid;
sum_entry->version = 0;
sum_entry->ofs_in_node = 0;
sum_entry++;
f2fs_put_page(page, 1);
}
invalidate_mapping_pages(META_MAPPING(sbi), addr,
addr + nrpages);
}
return 0;
}
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
static void remove_nats_in_journal(struct f2fs_sb_info *sbi)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA);
struct f2fs_journal *journal = curseg->journal;
int i;
down_write(&curseg->journal_rwsem);
for (i = 0; i < nats_in_cursum(journal); i++) {
struct nat_entry *ne;
struct f2fs_nat_entry raw_ne;
nid_t nid = le32_to_cpu(nid_in_journal(journal, i));
raw_ne = nat_in_journal(journal, i);
ne = __lookup_nat_cache(nm_i, nid);
if (!ne) {
ne = __alloc_nat_entry(nid, true);
__init_nat_entry(nm_i, ne, &raw_ne, true);
}
/*
* if a free nat in journal has not been used after last
* checkpoint, we should remove it from available nids,
* since later we will add it again.
*/
if (!get_nat_flag(ne, IS_DIRTY) &&
le32_to_cpu(raw_ne.block_addr) == NULL_ADDR) {
spin_lock(&nm_i->nid_list_lock);
nm_i->available_nids--;
spin_unlock(&nm_i->nid_list_lock);
}
__set_nat_cache_dirty(nm_i, ne);
}
update_nats_in_cursum(journal, -i);
up_write(&curseg->journal_rwsem);
}
static void __adjust_nat_entry_set(struct nat_entry_set *nes,
struct list_head *head, int max)
{
struct nat_entry_set *cur;
if (nes->entry_cnt >= max)
goto add_out;
list_for_each_entry(cur, head, set_list) {
if (cur->entry_cnt >= nes->entry_cnt) {
list_add(&nes->set_list, cur->set_list.prev);
return;
}
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
}
add_out:
list_add_tail(&nes->set_list, head);
}
static void __update_nat_bits(struct f2fs_sb_info *sbi, nid_t start_nid,
struct page *page)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
unsigned int nat_index = start_nid / NAT_ENTRY_PER_BLOCK;
struct f2fs_nat_block *nat_blk = page_address(page);
int valid = 0;
int i = 0;
if (!enabled_nat_bits(sbi, NULL))
return;
if (nat_index == 0) {
valid = 1;
i = 1;
}
for (; i < NAT_ENTRY_PER_BLOCK; i++) {
if (le32_to_cpu(nat_blk->entries[i].block_addr) != NULL_ADDR)
valid++;
}
if (valid == 0) {
__set_bit_le(nat_index, nm_i->empty_nat_bits);
__clear_bit_le(nat_index, nm_i->full_nat_bits);
return;
}
__clear_bit_le(nat_index, nm_i->empty_nat_bits);
if (valid == NAT_ENTRY_PER_BLOCK)
__set_bit_le(nat_index, nm_i->full_nat_bits);
else
__clear_bit_le(nat_index, nm_i->full_nat_bits);
}
static int __flush_nat_entry_set(struct f2fs_sb_info *sbi,
struct nat_entry_set *set, struct cp_control *cpc)
{
struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA);
struct f2fs_journal *journal = curseg->journal;
nid_t start_nid = set->set * NAT_ENTRY_PER_BLOCK;
bool to_journal = true;
struct f2fs_nat_block *nat_blk;
struct nat_entry *ne, *cur;
struct page *page = NULL;
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
/*
* there are two steps to flush nat entries:
* #1, flush nat entries to journal in current hot data summary block.
* #2, flush nat entries to nat page.
*/
if (enabled_nat_bits(sbi, cpc) ||
!__has_cursum_space(journal, set->entry_cnt, NAT_JOURNAL))
to_journal = false;
if (to_journal) {
down_write(&curseg->journal_rwsem);
} else {
page = get_next_nat_page(sbi, start_nid);
if (IS_ERR(page))
return PTR_ERR(page);
nat_blk = page_address(page);
f2fs_bug_on(sbi, !nat_blk);
}
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
/* flush dirty nats in nat entry set */
list_for_each_entry_safe(ne, cur, &set->entry_list, list) {
struct f2fs_nat_entry *raw_ne;
nid_t nid = nat_get_nid(ne);
int offset;
f2fs_bug_on(sbi, nat_get_blkaddr(ne) == NEW_ADDR);
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +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,
NAT_JOURNAL, nid, 1);
f2fs_bug_on(sbi, offset < 0);
raw_ne = &nat_in_journal(journal, offset);
nid_in_journal(journal, offset) = cpu_to_le32(nid);
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
} else {
raw_ne = &nat_blk->entries[nid - start_nid];
}
raw_nat_from_node_info(raw_ne, &ne->ni);
nat_reset_flag(ne);
__clear_nat_cache_dirty(NM_I(sbi), set, ne);
if (nat_get_blkaddr(ne) == NULL_ADDR) {
add_free_nid(sbi, nid, false, true);
} else {
spin_lock(&NM_I(sbi)->nid_list_lock);
update_free_nid_bitmap(sbi, nid, false, false);
spin_unlock(&NM_I(sbi)->nid_list_lock);
}
}
if (to_journal) {
up_write(&curseg->journal_rwsem);
} else {
__update_nat_bits(sbi, start_nid, page);
f2fs_put_page(page, 1);
}
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
/* Allow dirty nats by node block allocation in write_begin */
if (!set->entry_cnt) {
radix_tree_delete(&NM_I(sbi)->nat_set_root, set->set);
kmem_cache_free(nat_entry_set_slab, set);
}
return 0;
}
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
/*
* This function is called during the checkpointing process.
*/
int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA);
struct f2fs_journal *journal = curseg->journal;
struct nat_entry_set *setvec[SETVEC_SIZE];
struct nat_entry_set *set, *tmp;
unsigned int found;
nid_t set_idx = 0;
LIST_HEAD(sets);
int err = 0;
/* during unmount, let's flush nat_bits before checking dirty_nat_cnt */
if (enabled_nat_bits(sbi, cpc)) {
down_write(&nm_i->nat_tree_lock);
remove_nats_in_journal(sbi);
up_write(&nm_i->nat_tree_lock);
}
if (!nm_i->dirty_nat_cnt)
return 0;
down_write(&nm_i->nat_tree_lock);
/*
* if there are no enough space in journal to store dirty nat
* entries, remove all entries from journal and merge them
* into nat entry set.
*/
if (enabled_nat_bits(sbi, cpc) ||
!__has_cursum_space(journal, nm_i->dirty_nat_cnt, NAT_JOURNAL))
remove_nats_in_journal(sbi);
while ((found = __gang_lookup_nat_set(nm_i,
set_idx, SETVEC_SIZE, setvec))) {
unsigned idx;
set_idx = setvec[found - 1]->set + 1;
for (idx = 0; idx < found; idx++)
__adjust_nat_entry_set(setvec[idx], &sets,
MAX_NAT_JENTRIES(journal));
}
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
/* flush dirty nats in nat entry set */
list_for_each_entry_safe(set, tmp, &sets, set_list) {
err = __flush_nat_entry_set(sbi, set, cpc);
if (err)
break;
}
up_write(&nm_i->nat_tree_lock);
/* Allow dirty nats by node block allocation in write_begin */
return err;
}
static int __get_nat_bitmaps(struct f2fs_sb_info *sbi)
{
struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
struct f2fs_nm_info *nm_i = NM_I(sbi);
unsigned int nat_bits_bytes = nm_i->nat_blocks / BITS_PER_BYTE;
unsigned int i;
__u64 cp_ver = cur_cp_version(ckpt);
block_t nat_bits_addr;
if (!enabled_nat_bits(sbi, NULL))
return 0;
nm_i->nat_bits_blocks = F2FS_BLK_ALIGN((nat_bits_bytes << 1) + 8);
nm_i->nat_bits = f2fs_kzalloc(sbi,
nm_i->nat_bits_blocks << F2FS_BLKSIZE_BITS, GFP_KERNEL);
if (!nm_i->nat_bits)
return -ENOMEM;
nat_bits_addr = __start_cp_addr(sbi) + sbi->blocks_per_seg -
nm_i->nat_bits_blocks;
for (i = 0; i < nm_i->nat_bits_blocks; i++) {
struct page *page;
page = f2fs_get_meta_page(sbi, nat_bits_addr++);
if (IS_ERR(page))
return PTR_ERR(page);
memcpy(nm_i->nat_bits + (i << F2FS_BLKSIZE_BITS),
page_address(page), F2FS_BLKSIZE);
f2fs_put_page(page, 1);
}
cp_ver |= (cur_cp_crc(ckpt) << 32);
if (cpu_to_le64(cp_ver) != *(__le64 *)nm_i->nat_bits) {
disable_nat_bits(sbi, true);
return 0;
}
nm_i->full_nat_bits = nm_i->nat_bits + 8;
nm_i->empty_nat_bits = nm_i->full_nat_bits + nat_bits_bytes;
f2fs_notice(sbi, "Found nat_bits in checkpoint");
return 0;
}
static inline void load_free_nid_bitmap(struct f2fs_sb_info *sbi)
f2fs: combine nat_bits and free_nid_bitmap cache Both nat_bits cache and free_nid_bitmap cache provide same functionality as a intermediate cache between free nid cache and disk, but with different granularity of indicating free nid range, and different persistence policy. nat_bits cache provides better persistence ability, and free_nid_bitmap provides better granularity. In this patch we combine advantage of both caches, so finally policy of the intermediate cache would be: - init: load free nid status from nat_bits into free_nid_bitmap - lookup: scan free_nid_bitmap before load NAT blocks - update: update free_nid_bitmap in real-time - persistence: udpate and persist nat_bits in checkpoint This patch also resolves performance regression reported by lkp-robot. commit: 4ac912427c4214d8031d9ad6fbc3bc75e71512df ("f2fs: introduce free nid bitmap") d00030cf9cd0bb96fdccc41e33d3c91dcbb672ba ("f2fs: use __set{__clear}_bit_le") 1382c0f3f9d3f936c8bc42ed1591cf7a593ef9f7 ("f2fs: combine nat_bits and free_nid_bitmap cache") 4ac912427c4214d8 d00030cf9cd0bb96fdccc41e33 1382c0f3f9d3f936c8bc42ed15 ---------------- -------------------------- -------------------------- %stddev %change %stddev %change %stddev \ | \ | \ 77863 ± 0% +2.1% 79485 ± 1% +50.8% 117404 ± 0% aim7.jobs-per-min 231.63 ± 0% -2.0% 227.01 ± 1% -33.6% 153.80 ± 0% aim7.time.elapsed_time 231.63 ± 0% -2.0% 227.01 ± 1% -33.6% 153.80 ± 0% aim7.time.elapsed_time.max 896604 ± 0% -0.8% 889221 ± 3% -20.2% 715260 ± 1% aim7.time.involuntary_context_switches 2394 ± 1% +4.6% 2503 ± 1% +3.7% 2481 ± 2% aim7.time.maximum_resident_set_size 6240 ± 0% -1.5% 6145 ± 1% -14.1% 5360 ± 1% aim7.time.system_time 1111357 ± 3% +1.9% 1132509 ± 2% -6.2% 1041932 ± 2% aim7.time.voluntary_context_switches ... Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Xiaolong Ye <xiaolong.ye@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-03-08 12:07:49 +00:00
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
unsigned int i = 0;
nid_t nid, last_nid;
if (!enabled_nat_bits(sbi, NULL))
return;
for (i = 0; i < nm_i->nat_blocks; i++) {
i = find_next_bit_le(nm_i->empty_nat_bits, nm_i->nat_blocks, i);
if (i >= nm_i->nat_blocks)
break;
__set_bit_le(i, nm_i->nat_block_bitmap);
nid = i * NAT_ENTRY_PER_BLOCK;
last_nid = nid + NAT_ENTRY_PER_BLOCK;
f2fs: combine nat_bits and free_nid_bitmap cache Both nat_bits cache and free_nid_bitmap cache provide same functionality as a intermediate cache between free nid cache and disk, but with different granularity of indicating free nid range, and different persistence policy. nat_bits cache provides better persistence ability, and free_nid_bitmap provides better granularity. In this patch we combine advantage of both caches, so finally policy of the intermediate cache would be: - init: load free nid status from nat_bits into free_nid_bitmap - lookup: scan free_nid_bitmap before load NAT blocks - update: update free_nid_bitmap in real-time - persistence: udpate and persist nat_bits in checkpoint This patch also resolves performance regression reported by lkp-robot. commit: 4ac912427c4214d8031d9ad6fbc3bc75e71512df ("f2fs: introduce free nid bitmap") d00030cf9cd0bb96fdccc41e33d3c91dcbb672ba ("f2fs: use __set{__clear}_bit_le") 1382c0f3f9d3f936c8bc42ed1591cf7a593ef9f7 ("f2fs: combine nat_bits and free_nid_bitmap cache") 4ac912427c4214d8 d00030cf9cd0bb96fdccc41e33 1382c0f3f9d3f936c8bc42ed15 ---------------- -------------------------- -------------------------- %stddev %change %stddev %change %stddev \ | \ | \ 77863 ± 0% +2.1% 79485 ± 1% +50.8% 117404 ± 0% aim7.jobs-per-min 231.63 ± 0% -2.0% 227.01 ± 1% -33.6% 153.80 ± 0% aim7.time.elapsed_time 231.63 ± 0% -2.0% 227.01 ± 1% -33.6% 153.80 ± 0% aim7.time.elapsed_time.max 896604 ± 0% -0.8% 889221 ± 3% -20.2% 715260 ± 1% aim7.time.involuntary_context_switches 2394 ± 1% +4.6% 2503 ± 1% +3.7% 2481 ± 2% aim7.time.maximum_resident_set_size 6240 ± 0% -1.5% 6145 ± 1% -14.1% 5360 ± 1% aim7.time.system_time 1111357 ± 3% +1.9% 1132509 ± 2% -6.2% 1041932 ± 2% aim7.time.voluntary_context_switches ... Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Xiaolong Ye <xiaolong.ye@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-03-08 12:07:49 +00:00
spin_lock(&NM_I(sbi)->nid_list_lock);
f2fs: combine nat_bits and free_nid_bitmap cache Both nat_bits cache and free_nid_bitmap cache provide same functionality as a intermediate cache between free nid cache and disk, but with different granularity of indicating free nid range, and different persistence policy. nat_bits cache provides better persistence ability, and free_nid_bitmap provides better granularity. In this patch we combine advantage of both caches, so finally policy of the intermediate cache would be: - init: load free nid status from nat_bits into free_nid_bitmap - lookup: scan free_nid_bitmap before load NAT blocks - update: update free_nid_bitmap in real-time - persistence: udpate and persist nat_bits in checkpoint This patch also resolves performance regression reported by lkp-robot. commit: 4ac912427c4214d8031d9ad6fbc3bc75e71512df ("f2fs: introduce free nid bitmap") d00030cf9cd0bb96fdccc41e33d3c91dcbb672ba ("f2fs: use __set{__clear}_bit_le") 1382c0f3f9d3f936c8bc42ed1591cf7a593ef9f7 ("f2fs: combine nat_bits and free_nid_bitmap cache") 4ac912427c4214d8 d00030cf9cd0bb96fdccc41e33 1382c0f3f9d3f936c8bc42ed15 ---------------- -------------------------- -------------------------- %stddev %change %stddev %change %stddev \ | \ | \ 77863 ± 0% +2.1% 79485 ± 1% +50.8% 117404 ± 0% aim7.jobs-per-min 231.63 ± 0% -2.0% 227.01 ± 1% -33.6% 153.80 ± 0% aim7.time.elapsed_time 231.63 ± 0% -2.0% 227.01 ± 1% -33.6% 153.80 ± 0% aim7.time.elapsed_time.max 896604 ± 0% -0.8% 889221 ± 3% -20.2% 715260 ± 1% aim7.time.involuntary_context_switches 2394 ± 1% +4.6% 2503 ± 1% +3.7% 2481 ± 2% aim7.time.maximum_resident_set_size 6240 ± 0% -1.5% 6145 ± 1% -14.1% 5360 ± 1% aim7.time.system_time 1111357 ± 3% +1.9% 1132509 ± 2% -6.2% 1041932 ± 2% aim7.time.voluntary_context_switches ... Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Xiaolong Ye <xiaolong.ye@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-03-08 12:07:49 +00:00
for (; nid < last_nid; nid++)
update_free_nid_bitmap(sbi, nid, true, true);
spin_unlock(&NM_I(sbi)->nid_list_lock);
f2fs: combine nat_bits and free_nid_bitmap cache Both nat_bits cache and free_nid_bitmap cache provide same functionality as a intermediate cache between free nid cache and disk, but with different granularity of indicating free nid range, and different persistence policy. nat_bits cache provides better persistence ability, and free_nid_bitmap provides better granularity. In this patch we combine advantage of both caches, so finally policy of the intermediate cache would be: - init: load free nid status from nat_bits into free_nid_bitmap - lookup: scan free_nid_bitmap before load NAT blocks - update: update free_nid_bitmap in real-time - persistence: udpate and persist nat_bits in checkpoint This patch also resolves performance regression reported by lkp-robot. commit: 4ac912427c4214d8031d9ad6fbc3bc75e71512df ("f2fs: introduce free nid bitmap") d00030cf9cd0bb96fdccc41e33d3c91dcbb672ba ("f2fs: use __set{__clear}_bit_le") 1382c0f3f9d3f936c8bc42ed1591cf7a593ef9f7 ("f2fs: combine nat_bits and free_nid_bitmap cache") 4ac912427c4214d8 d00030cf9cd0bb96fdccc41e33 1382c0f3f9d3f936c8bc42ed15 ---------------- -------------------------- -------------------------- %stddev %change %stddev %change %stddev \ | \ | \ 77863 ± 0% +2.1% 79485 ± 1% +50.8% 117404 ± 0% aim7.jobs-per-min 231.63 ± 0% -2.0% 227.01 ± 1% -33.6% 153.80 ± 0% aim7.time.elapsed_time 231.63 ± 0% -2.0% 227.01 ± 1% -33.6% 153.80 ± 0% aim7.time.elapsed_time.max 896604 ± 0% -0.8% 889221 ± 3% -20.2% 715260 ± 1% aim7.time.involuntary_context_switches 2394 ± 1% +4.6% 2503 ± 1% +3.7% 2481 ± 2% aim7.time.maximum_resident_set_size 6240 ± 0% -1.5% 6145 ± 1% -14.1% 5360 ± 1% aim7.time.system_time 1111357 ± 3% +1.9% 1132509 ± 2% -6.2% 1041932 ± 2% aim7.time.voluntary_context_switches ... Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Xiaolong Ye <xiaolong.ye@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-03-08 12:07:49 +00:00
}
for (i = 0; i < nm_i->nat_blocks; i++) {
i = find_next_bit_le(nm_i->full_nat_bits, nm_i->nat_blocks, i);
if (i >= nm_i->nat_blocks)
break;
__set_bit_le(i, nm_i->nat_block_bitmap);
}
}
static int init_node_manager(struct f2fs_sb_info *sbi)
{
struct f2fs_super_block *sb_raw = F2FS_RAW_SUPER(sbi);
struct f2fs_nm_info *nm_i = NM_I(sbi);
unsigned char *version_bitmap;
unsigned int nat_segs;
int err;
nm_i->nat_blkaddr = le32_to_cpu(sb_raw->nat_blkaddr);
/* segment_count_nat includes pair segment so divide to 2. */
nat_segs = le32_to_cpu(sb_raw->segment_count_nat) >> 1;
nm_i->nat_blocks = nat_segs << le32_to_cpu(sb_raw->log_blocks_per_seg);
nm_i->max_nid = NAT_ENTRY_PER_BLOCK * nm_i->nat_blocks;
/* not used nids: 0, node, meta, (and root counted as valid node) */
nm_i->available_nids = nm_i->max_nid - sbi->total_valid_node_count -
F2FS_RESERVED_NODE_NUM;
nm_i->nid_cnt[FREE_NID] = 0;
nm_i->nid_cnt[PREALLOC_NID] = 0;
nm_i->nat_cnt = 0;
nm_i->ram_thresh = DEF_RAM_THRESHOLD;
nm_i->ra_nid_pages = DEF_RA_NID_PAGES;
nm_i->dirty_nats_ratio = DEF_DIRTY_NAT_RATIO_THRESHOLD;
INIT_RADIX_TREE(&nm_i->free_nid_root, GFP_ATOMIC);
INIT_LIST_HEAD(&nm_i->free_nid_list);
INIT_RADIX_TREE(&nm_i->nat_root, GFP_NOIO);
INIT_RADIX_TREE(&nm_i->nat_set_root, GFP_NOIO);
INIT_LIST_HEAD(&nm_i->nat_entries);
spin_lock_init(&nm_i->nat_list_lock);
mutex_init(&nm_i->build_lock);
spin_lock_init(&nm_i->nid_list_lock);
init_rwsem(&nm_i->nat_tree_lock);
nm_i->next_scan_nid = le32_to_cpu(sbi->ckpt->next_free_nid);
nm_i->bitmap_size = __bitmap_size(sbi, NAT_BITMAP);
version_bitmap = __bitmap_ptr(sbi, NAT_BITMAP);
if (!version_bitmap)
return -EFAULT;
nm_i->nat_bitmap = kmemdup(version_bitmap, nm_i->bitmap_size,
GFP_KERNEL);
if (!nm_i->nat_bitmap)
return -ENOMEM;
err = __get_nat_bitmaps(sbi);
if (err)
return err;
#ifdef CONFIG_F2FS_CHECK_FS
nm_i->nat_bitmap_mir = kmemdup(version_bitmap, nm_i->bitmap_size,
GFP_KERNEL);
if (!nm_i->nat_bitmap_mir)
return -ENOMEM;
#endif
return 0;
}
static int init_free_nid_cache(struct f2fs_sb_info *sbi)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
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
nm_i->free_nid_bitmap =
f2fs_kzalloc(sbi, array_size(sizeof(unsigned char *),
nm_i->nat_blocks),
GFP_KERNEL);
if (!nm_i->free_nid_bitmap)
return -ENOMEM;
for (i = 0; i < nm_i->nat_blocks; i++) {
nm_i->free_nid_bitmap[i] = f2fs_kvzalloc(sbi,
f2fs_bitmap_size(NAT_ENTRY_PER_BLOCK), GFP_KERNEL);
if (!nm_i->free_nid_bitmap[i])
return -ENOMEM;
}
nm_i->nat_block_bitmap = f2fs_kvzalloc(sbi, nm_i->nat_blocks / 8,
GFP_KERNEL);
if (!nm_i->nat_block_bitmap)
return -ENOMEM;
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
nm_i->free_nid_count =
f2fs_kvzalloc(sbi, array_size(sizeof(unsigned short),
nm_i->nat_blocks),
GFP_KERNEL);
if (!nm_i->free_nid_count)
return -ENOMEM;
return 0;
}
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_build_node_manager(struct f2fs_sb_info *sbi)
{
int err;
sbi->nm_info = f2fs_kzalloc(sbi, sizeof(struct f2fs_nm_info),
GFP_KERNEL);
if (!sbi->nm_info)
return -ENOMEM;
err = init_node_manager(sbi);
if (err)
return err;
err = init_free_nid_cache(sbi);
if (err)
return err;
f2fs: combine nat_bits and free_nid_bitmap cache Both nat_bits cache and free_nid_bitmap cache provide same functionality as a intermediate cache between free nid cache and disk, but with different granularity of indicating free nid range, and different persistence policy. nat_bits cache provides better persistence ability, and free_nid_bitmap provides better granularity. In this patch we combine advantage of both caches, so finally policy of the intermediate cache would be: - init: load free nid status from nat_bits into free_nid_bitmap - lookup: scan free_nid_bitmap before load NAT blocks - update: update free_nid_bitmap in real-time - persistence: udpate and persist nat_bits in checkpoint This patch also resolves performance regression reported by lkp-robot. commit: 4ac912427c4214d8031d9ad6fbc3bc75e71512df ("f2fs: introduce free nid bitmap") d00030cf9cd0bb96fdccc41e33d3c91dcbb672ba ("f2fs: use __set{__clear}_bit_le") 1382c0f3f9d3f936c8bc42ed1591cf7a593ef9f7 ("f2fs: combine nat_bits and free_nid_bitmap cache") 4ac912427c4214d8 d00030cf9cd0bb96fdccc41e33 1382c0f3f9d3f936c8bc42ed15 ---------------- -------------------------- -------------------------- %stddev %change %stddev %change %stddev \ | \ | \ 77863 ± 0% +2.1% 79485 ± 1% +50.8% 117404 ± 0% aim7.jobs-per-min 231.63 ± 0% -2.0% 227.01 ± 1% -33.6% 153.80 ± 0% aim7.time.elapsed_time 231.63 ± 0% -2.0% 227.01 ± 1% -33.6% 153.80 ± 0% aim7.time.elapsed_time.max 896604 ± 0% -0.8% 889221 ± 3% -20.2% 715260 ± 1% aim7.time.involuntary_context_switches 2394 ± 1% +4.6% 2503 ± 1% +3.7% 2481 ± 2% aim7.time.maximum_resident_set_size 6240 ± 0% -1.5% 6145 ± 1% -14.1% 5360 ± 1% aim7.time.system_time 1111357 ± 3% +1.9% 1132509 ± 2% -6.2% 1041932 ± 2% aim7.time.voluntary_context_switches ... Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Xiaolong Ye <xiaolong.ye@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-03-08 12:07:49 +00:00
/* load free nid status from nat_bits table */
load_free_nid_bitmap(sbi);
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
return f2fs_build_free_nids(sbi, true, true);
}
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
struct free_nid *i, *next_i;
struct nat_entry *natvec[NATVEC_SIZE];
struct nat_entry_set *setvec[SETVEC_SIZE];
nid_t nid = 0;
unsigned int found;
if (!nm_i)
return;
/* destroy free nid list */
spin_lock(&nm_i->nid_list_lock);
list_for_each_entry_safe(i, next_i, &nm_i->free_nid_list, list) {
__remove_free_nid(sbi, i, FREE_NID);
spin_unlock(&nm_i->nid_list_lock);
kmem_cache_free(free_nid_slab, i);
spin_lock(&nm_i->nid_list_lock);
}
f2fs_bug_on(sbi, nm_i->nid_cnt[FREE_NID]);
f2fs_bug_on(sbi, nm_i->nid_cnt[PREALLOC_NID]);
f2fs_bug_on(sbi, !list_empty(&nm_i->free_nid_list));
spin_unlock(&nm_i->nid_list_lock);
/* destroy nat cache */
down_write(&nm_i->nat_tree_lock);
while ((found = __gang_lookup_nat_cache(nm_i,
nid, NATVEC_SIZE, natvec))) {
unsigned idx;
nid = nat_get_nid(natvec[found - 1]) + 1;
for (idx = 0; idx < found; idx++) {
spin_lock(&nm_i->nat_list_lock);
list_del(&natvec[idx]->list);
spin_unlock(&nm_i->nat_list_lock);
__del_from_nat_cache(nm_i, natvec[idx]);
}
}
f2fs_bug_on(sbi, nm_i->nat_cnt);
/* destroy nat set cache */
nid = 0;
while ((found = __gang_lookup_nat_set(nm_i,
nid, SETVEC_SIZE, setvec))) {
unsigned idx;
nid = setvec[found - 1]->set + 1;
for (idx = 0; idx < found; idx++) {
/* entry_cnt is not zero, when cp_error was occurred */
f2fs_bug_on(sbi, !list_empty(&setvec[idx]->entry_list));
radix_tree_delete(&nm_i->nat_set_root, setvec[idx]->set);
kmem_cache_free(nat_entry_set_slab, setvec[idx]);
}
}
up_write(&nm_i->nat_tree_lock);
kvfree(nm_i->nat_block_bitmap);
if (nm_i->free_nid_bitmap) {
int i;
for (i = 0; i < nm_i->nat_blocks; i++)
kvfree(nm_i->free_nid_bitmap[i]);
kvfree(nm_i->free_nid_bitmap);
}
kvfree(nm_i->free_nid_count);
kvfree(nm_i->nat_bitmap);
kvfree(nm_i->nat_bits);
#ifdef CONFIG_F2FS_CHECK_FS
kvfree(nm_i->nat_bitmap_mir);
#endif
sbi->nm_info = NULL;
kvfree(nm_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
int __init f2fs_create_node_manager_caches(void)
{
nat_entry_slab = f2fs_kmem_cache_create("nat_entry",
sizeof(struct nat_entry));
if (!nat_entry_slab)
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
goto fail;
free_nid_slab = f2fs_kmem_cache_create("free_nid",
sizeof(struct free_nid));
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
if (!free_nid_slab)
goto destroy_nat_entry;
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
nat_entry_set_slab = f2fs_kmem_cache_create("nat_entry_set",
sizeof(struct nat_entry_set));
if (!nat_entry_set_slab)
goto destroy_free_nid;
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
fsync_node_entry_slab = f2fs_kmem_cache_create("fsync_node_entry",
sizeof(struct fsync_node_entry));
if (!fsync_node_entry_slab)
goto destroy_nat_entry_set;
return 0;
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
destroy_nat_entry_set:
kmem_cache_destroy(nat_entry_set_slab);
destroy_free_nid:
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
kmem_cache_destroy(free_nid_slab);
destroy_nat_entry:
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
kmem_cache_destroy(nat_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_node_manager_caches(void)
{
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
kmem_cache_destroy(fsync_node_entry_slab);
f2fs: refactor flush_nat_entries codes for reducing NAT writes 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. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
kmem_cache_destroy(nat_entry_set_slab);
kmem_cache_destroy(free_nid_slab);
kmem_cache_destroy(nat_entry_slab);
}