p71924506
/
linux_old1
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
linux_old1/fs/f2fs/recovery.c

553 lines
13 KiB
C
Raw Normal View History

f2fs: adjust kernel coding style As pointed out by Randy Dunlap, this patch removes all usage of "/**" for comment blocks. Instead, just use "/*". Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-29 12:28:09 +08:00
/*
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
* fs/f2fs/recovery.c
*
* Copyright (c) 2012 Samsung Electronics Co., Ltd.
* http://www.samsung.com/
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
*/
#include <linux/fs.h>
#include <linux/f2fs_fs.h>
#include "f2fs.h"
#include "node.h"
#include "segment.h"
f2fs: fix roll-forward missing scenarios We can summarize the roll forward recovery scenarios as follows. [Term] F: fsync_mark, D: dentry_mark 1. inode(x) | CP | inode(x) | dnode(F) -> Update the latest inode(x). 2. inode(x) | CP | inode(F) | dnode(F) -> No problem. 3. inode(x) | CP | dnode(F) | inode(x) -> Recover to the latest dnode(F), and drop the last inode(x) 4. inode(x) | CP | dnode(F) | inode(F) -> No problem. 5. CP | inode(x) | dnode(F) -> The inode(DF) was missing. Should drop this dnode(F). 6. CP | inode(DF) | dnode(F) -> No problem. 7. CP | dnode(F) | inode(DF) -> If f2fs_iget fails, then goto next to find inode(DF). 8. CP | dnode(F) | inode(x) -> If f2fs_iget fails, then goto next to find inode(DF). But it will fail due to no inode(DF). So, this patch adds some missing points such as #1, #5, #7, and #8. Signed-off-by: Huang Ying <ying.huang@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-16 07:46:08 +08:00
/*
* Roll forward recovery scenarios.
*
* [Term] F: fsync_mark, D: dentry_mark
*
* 1. inode(x) | CP | inode(x) | dnode(F)
* -> Update the latest inode(x).
*
* 2. inode(x) | CP | inode(F) | dnode(F)
* -> No problem.
*
* 3. inode(x) | CP | dnode(F) | inode(x)
* -> Recover to the latest dnode(F), and drop the last inode(x)
*
* 4. inode(x) | CP | dnode(F) | inode(F)
* -> No problem.
*
* 5. CP | inode(x) | dnode(F)
* -> The inode(DF) was missing. Should drop this dnode(F).
*
* 6. CP | inode(DF) | dnode(F)
* -> No problem.
*
* 7. CP | dnode(F) | inode(DF)
* -> If f2fs_iget fails, then goto next to find inode(DF).
*
* 8. CP | dnode(F) | inode(x)
* -> If f2fs_iget fails, then goto next to find inode(DF).
* But it will fail due to no inode(DF).
*/
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
static struct kmem_cache *fsync_entry_slab;
bool space_for_roll_forward(struct f2fs_sb_info *sbi)
{
if (sbi->last_valid_block_count + sbi->alloc_valid_block_count
> sbi->user_block_count)
return false;
return true;
}
static struct fsync_inode_entry *get_fsync_inode(struct list_head *head,
nid_t ino)
{
struct fsync_inode_entry *entry;
f2fs: use list_for_each_entry{_safe} for simplyfying code This patch use list_for_each_entry{_safe} instead of list_for_each{_safe} for simplfying code. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-03-29 11:33:17 +08:00
list_for_each_entry(entry, head, list)
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
if (entry->inode->i_ino == ino)
return entry;
f2fs: use list_for_each_entry{_safe} for simplyfying code This patch use list_for_each_entry{_safe} instead of list_for_each{_safe} for simplfying code. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-03-29 11:33:17 +08:00
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
return NULL;
}
f2fs: remove redundant operation during roll-forward recovery If same data is updated multiple times, we don't need to redo whole the operations. Let's just update the lastest one. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 05:29:06 +08:00
static int recover_dentry(struct inode *inode, struct page *ipage)
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
{
f2fs: introduce F2FS_INODE macro to get f2fs_inode This patch introduces F2FS_INODE that returns struct f2fs_inode * from the inode page. By using this macro, we can remove unnecessary casting codes like below. struct f2fs_inode *ri = &F2FS_NODE(inode_page)->i; -> struct f2fs_inode *ri = F2FS_INODE(inode_page); Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-12-26 15:30:41 +08:00
struct f2fs_inode *raw_inode = F2FS_INODE(ipage);
f2fs: fix BUG_ON during f2fs_evict_inode(dir) During the dentry recovery routine, recover_inode() triggers __f2fs_add_link with its directory inode. In the following scenario, a bug is captured. 1. dir = f2fs_iget(pino) 2. __f2fs_add_link(dir, name) 3. iput(dir) -> f2fs_evict_inode() faces with BUG_ON(atomic_read(fi->dirty_dents)) Kernel BUG at ffffffffa01c0676 [verbose debug info unavailable] [<ffffffffa01c0676>] f2fs_evict_inode+0x276/0x300 [f2fs] Call Trace: [<ffffffff8118ea00>] evict+0xb0/0x1b0 [<ffffffff8118f1c5>] iput+0x105/0x190 [<ffffffffa01d2dac>] recover_fsync_data+0x3bc/0x1070 [f2fs] [<ffffffff81692e8a>] ? io_schedule+0xaa/0xd0 [<ffffffff81690acb>] ? __wait_on_bit_lock+0x7b/0xc0 [<ffffffff8111a0e7>] ? __lock_page+0x67/0x70 [<ffffffff81165e21>] ? kmem_cache_alloc+0x31/0x140 [<ffffffff8118a502>] ? __d_instantiate+0x92/0xf0 [<ffffffff812a949b>] ? security_d_instantiate+0x1b/0x30 [<ffffffff8118a5b4>] ? d_instantiate+0x54/0x70 This means that we should flush all the dentry pages between iget and iput(). But, during the recovery routine, it is unallowed due to consistency, so we have to wait the whole recovery process. And then, write_checkpoint flushes all the dirty dentry blocks, and nicely we can put the stale dir inodes from the dirty_dir_inode_list. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-15 15:40:02 +08:00
nid_t pino = le32_to_cpu(raw_inode->i_pino);
f2fs: fix dentry recovery routine The error scenario is: 1. create /a (1.a link /a /b) 2. sync 3. unlinke /a 4. create /a 5. fsync /a 6. Sudden power-off When the f2fs recovers the fsynced dentry, /a, we discover an exsiting dentry at f2fs_find_entry() in recover_dentry(). In such the case, we should unlink the existing dentry and its inode and then recover newly created dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-28 08:19:22 +08:00
struct f2fs_dir_entry *de;
f2fs: get rid of fake on-stack dentries those should never be used for a lot of reasons... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2013-01-26 05:15:43 +08:00
struct qstr name;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
struct page *page;
f2fs: fix dentry recovery routine The error scenario is: 1. create /a (1.a link /a /b) 2. sync 3. unlinke /a 4. create /a 5. fsync /a 6. Sudden power-off When the f2fs recovers the fsynced dentry, /a, we discover an exsiting dentry at f2fs_find_entry() in recover_dentry(). In such the case, we should unlink the existing dentry and its inode and then recover newly created dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-28 08:19:22 +08:00
struct inode *dir, *einode;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
int err = 0;
f2fs: remove costly dirty_dir_inode operations This patch removes list opeations in handling dirty dir inodes. Previously, F2FS traverses whole the list of dirty dir inodes to check whether there is an existing inode or not, resulting in heavy CPU overheads. So this patch removes such the traverse operations by adding FI_DIRTY_DIR to indicate the inode lies on the list or not. Through this simple flag, we can remove redundant operations gracefully. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-04-15 10:19:28 +08:00
dir = f2fs_iget(inode->i_sb, pino);
if (IS_ERR(dir)) {
err = PTR_ERR(dir);
goto out;
}
f2fs: get rid of fake on-stack dentries those should never be used for a lot of reasons... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2013-01-26 05:15:43 +08:00
name.len = le32_to_cpu(raw_inode->i_namelen);
name.name = raw_inode->i_name;
f2fs: check filename length in recover_dentry In current flow, we will get Null return value of f2fs_find_entry in recover_dentry when name.len is bigger than F2FS_NAME_LEN, and then we still add this inode into its dir entry. To avoid this situation, we must check filename length before we use it. Another point is that we could remove the code of checking filename length In f2fs_find_entry, because f2fs_lookup will be called previously to ensure of validity of filename length. V2: o add WARN_ON() as Jaegeuk Kim suggested. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-12-23 11:12:21 +08:00
if (unlikely(name.len > F2FS_NAME_LEN)) {
WARN_ON(1);
err = -ENAMETOOLONG;
f2fs: avoid not to call remove_dirty_inode There is an errorneous case during the recovery like below. In recovery_dentry, 1) dir = f2fs_iget(); 2) mark the dir with FI_DELAY_IPUT 3) goto unmap_out After the end of recovery routine, there is no dirty dentries so the dir cannot be released by iput in remove_dirty_dir_inode. This patch fixes such the bug case by handling the iget and iput in the recovery_dentry procedure. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-07 02:05:03 +08:00
goto out_err;
f2fs: check filename length in recover_dentry In current flow, we will get Null return value of f2fs_find_entry in recover_dentry when name.len is bigger than F2FS_NAME_LEN, and then we still add this inode into its dir entry. To avoid this situation, we must check filename length before we use it. Another point is that we could remove the code of checking filename length In f2fs_find_entry, because f2fs_lookup will be called previously to ensure of validity of filename length. V2: o add WARN_ON() as Jaegeuk Kim suggested. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-12-23 11:12:21 +08:00
}
f2fs: fix dentry recovery routine The error scenario is: 1. create /a (1.a link /a /b) 2. sync 3. unlinke /a 4. create /a 5. fsync /a 6. Sudden power-off When the f2fs recovers the fsynced dentry, /a, we discover an exsiting dentry at f2fs_find_entry() in recover_dentry(). In such the case, we should unlink the existing dentry and its inode and then recover newly created dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-28 08:19:22 +08:00
retry:
de = f2fs_find_entry(dir, &name, &page);
f2fs: clear FI_INC_LINK during the recovery If an inode are fsynced multiple times with fsync & dent marks, this inode will set FI_INC_LINK at find_fsync_dnodes during the recovery. But, in recover_inode, recover_dentry doesn't clear that flag when multiple hits were occurred. So this patch removes the flag for the further consistency. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-08-08 08:06:18 +08:00
if (de && inode->i_ino == le32_to_cpu(de->ino)) {
clear_inode_flag(F2FS_I(inode), FI_INC_LINK);
f2fs: account for orphan inodes during recovery During recovery, orphan inodes are deleted via truncate_hole(). These orphans are added by recover_dentry() via f2fs_delete_entry(). However, f2fs_delete_entry() adds them via add_orphan_inode() without calling acquire_orphan_inode() first. This prevents the counters from being incremented properly, which causes them to underflow when remove_orphan_inode() is called later on. Signed-off-by: Russ Knize <rknize@motorola.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-09-24 22:40:57 +08:00
goto out_unmap_put;
f2fs: clear FI_INC_LINK during the recovery If an inode are fsynced multiple times with fsync & dent marks, this inode will set FI_INC_LINK at find_fsync_dnodes during the recovery. But, in recover_inode, recover_dentry doesn't clear that flag when multiple hits were occurred. So this patch removes the flag for the further consistency. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-08-08 08:06:18 +08:00
}
f2fs: fix dentry recovery routine The error scenario is: 1. create /a (1.a link /a /b) 2. sync 3. unlinke /a 4. create /a 5. fsync /a 6. Sudden power-off When the f2fs recovers the fsynced dentry, /a, we discover an exsiting dentry at f2fs_find_entry() in recover_dentry(). In such the case, we should unlink the existing dentry and its inode and then recover newly created dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-28 08:19:22 +08:00
if (de) {
einode = f2fs_iget(inode->i_sb, le32_to_cpu(de->ino));
if (IS_ERR(einode)) {
WARN_ON(1);
f2fs: set errno when f2fs_iget failed in recover_dentry We should set the error number correctly when we fail in recover_dentry(), so the recover flow could stop for the reason as error number shows instead of continuing. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-04-28 17:58:34 +08:00
err = PTR_ERR(einode);
if (err == -ENOENT)
f2fs: fix dentry recovery routine The error scenario is: 1. create /a (1.a link /a /b) 2. sync 3. unlinke /a 4. create /a 5. fsync /a 6. Sudden power-off When the f2fs recovers the fsynced dentry, /a, we discover an exsiting dentry at f2fs_find_entry() in recover_dentry(). In such the case, we should unlink the existing dentry and its inode and then recover newly created dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-28 08:19:22 +08:00
err = -EEXIST;
f2fs: account for orphan inodes during recovery During recovery, orphan inodes are deleted via truncate_hole(). These orphans are added by recover_dentry() via f2fs_delete_entry(). However, f2fs_delete_entry() adds them via add_orphan_inode() without calling acquire_orphan_inode() first. This prevents the counters from being incremented properly, which causes them to underflow when remove_orphan_inode() is called later on. Signed-off-by: Russ Knize <rknize@motorola.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-09-24 22:40:57 +08:00
goto out_unmap_put;
}
f2fs: introduce F2FS_I_SB, F2FS_M_SB, and F2FS_P_SB This patch adds three inline functions to clean up dirty casting codes. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-03 06:31:18 +08:00
err = acquire_orphan_inode(F2FS_I_SB(inode));
f2fs: account for orphan inodes during recovery During recovery, orphan inodes are deleted via truncate_hole(). These orphans are added by recover_dentry() via f2fs_delete_entry(). However, f2fs_delete_entry() adds them via add_orphan_inode() without calling acquire_orphan_inode() first. This prevents the counters from being incremented properly, which causes them to underflow when remove_orphan_inode() is called later on. Signed-off-by: Russ Knize <rknize@motorola.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-09-24 22:40:57 +08:00
if (err) {
iput(einode);
goto out_unmap_put;
f2fs: fix dentry recovery routine The error scenario is: 1. create /a (1.a link /a /b) 2. sync 3. unlinke /a 4. create /a 5. fsync /a 6. Sudden power-off When the f2fs recovers the fsynced dentry, /a, we discover an exsiting dentry at f2fs_find_entry() in recover_dentry(). In such the case, we should unlink the existing dentry and its inode and then recover newly created dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-28 08:19:22 +08:00
}
f2fs: export dir operations for inline dir This patch exports some dir operations for inline dir, additionally introduces f2fs_drop_nlink from f2fs_delete_entry for reusing by inline dir function. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-24 18:17:04 +08:00
f2fs_delete_entry(de, page, dir, einode);
f2fs: fix dentry recovery routine The error scenario is: 1. create /a (1.a link /a /b) 2. sync 3. unlinke /a 4. create /a 5. fsync /a 6. Sudden power-off When the f2fs recovers the fsynced dentry, /a, we discover an exsiting dentry at f2fs_find_entry() in recover_dentry(). In such the case, we should unlink the existing dentry and its inode and then recover newly created dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-28 08:19:22 +08:00
iput(einode);
goto retry;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
}
f2fs: fix dentry recovery routine The error scenario is: 1. create /a (1.a link /a /b) 2. sync 3. unlinke /a 4. create /a 5. fsync /a 6. Sudden power-off When the f2fs recovers the fsynced dentry, /a, we discover an exsiting dentry at f2fs_find_entry() in recover_dentry(). In such the case, we should unlink the existing dentry and its inode and then recover newly created dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-28 08:19:22 +08:00
err = __f2fs_add_link(dir, &name, inode);
f2fs: avoid not to call remove_dirty_inode There is an errorneous case during the recovery like below. In recovery_dentry, 1) dir = f2fs_iget(); 2) mark the dir with FI_DELAY_IPUT 3) goto unmap_out After the end of recovery routine, there is no dirty dentries so the dir cannot be released by iput in remove_dirty_dir_inode. This patch fixes such the bug case by handling the iget and iput in the recovery_dentry procedure. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-07 02:05:03 +08:00
if (err)
goto out_err;
if (is_inode_flag_set(F2FS_I(dir), FI_DELAY_IPUT)) {
iput(dir);
} else {
add_dirty_dir_inode(dir);
set_inode_flag(F2FS_I(dir), FI_DELAY_IPUT);
}
f2fs: account for orphan inodes during recovery During recovery, orphan inodes are deleted via truncate_hole(). These orphans are added by recover_dentry() via f2fs_delete_entry(). However, f2fs_delete_entry() adds them via add_orphan_inode() without calling acquire_orphan_inode() first. This prevents the counters from being incremented properly, which causes them to underflow when remove_orphan_inode() is called later on. Signed-off-by: Russ Knize <rknize@motorola.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-09-24 22:40:57 +08:00
goto out;
out_unmap_put:
f2fs: introduce f2fs_dentry_kunmap to clean up This patch introduces f2fs_dentry_kunmap to clean up dirty codes. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-11-22 08:36:28 +08:00
f2fs_dentry_kunmap(dir, page);
f2fs: account for orphan inodes during recovery During recovery, orphan inodes are deleted via truncate_hole(). These orphans are added by recover_dentry() via f2fs_delete_entry(). However, f2fs_delete_entry() adds them via add_orphan_inode() without calling acquire_orphan_inode() first. This prevents the counters from being incremented properly, which causes them to underflow when remove_orphan_inode() is called later on. Signed-off-by: Russ Knize <rknize@motorola.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-09-24 22:40:57 +08:00
f2fs_put_page(page, 0);
f2fs: avoid not to call remove_dirty_inode There is an errorneous case during the recovery like below. In recovery_dentry, 1) dir = f2fs_iget(); 2) mark the dir with FI_DELAY_IPUT 3) goto unmap_out After the end of recovery routine, there is no dirty dentries so the dir cannot be released by iput in remove_dirty_dir_inode. This patch fixes such the bug case by handling the iget and iput in the recovery_dentry procedure. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-07 02:05:03 +08:00
out_err:
iput(dir);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
out:
f2fs: clean checkpatch warnings Fixed a variety of trivial checkpatch warnings. The only delta should be some minor formatting on log strings that were split / too long. Signed-off-by: Chris Fries <cfries@motorola.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-01-18 04:44:39 +08:00
f2fs_msg(inode->i_sb, KERN_NOTICE,
"%s: ino = %x, name = %s, dir = %lx, err = %d",
__func__, ino_of_node(ipage), raw_inode->i_name,
f2fs: dereferencing an ERR_PTR There is an error path where "dir" is an ERR_PTR. Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-23 18:02:13 +08:00
IS_ERR(dir) ? 0 : dir->i_ino, err);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
return err;
}
f2fs: remove redundant operation during roll-forward recovery If same data is updated multiple times, we don't need to redo whole the operations. Let's just update the lastest one. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 05:29:06 +08:00
static void recover_inode(struct inode *inode, struct page *page)
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
{
f2fs: fix roll-forward missing scenarios We can summarize the roll forward recovery scenarios as follows. [Term] F: fsync_mark, D: dentry_mark 1. inode(x) | CP | inode(x) | dnode(F) -> Update the latest inode(x). 2. inode(x) | CP | inode(F) | dnode(F) -> No problem. 3. inode(x) | CP | dnode(F) | inode(x) -> Recover to the latest dnode(F), and drop the last inode(x) 4. inode(x) | CP | dnode(F) | inode(F) -> No problem. 5. CP | inode(x) | dnode(F) -> The inode(DF) was missing. Should drop this dnode(F). 6. CP | inode(DF) | dnode(F) -> No problem. 7. CP | dnode(F) | inode(DF) -> If f2fs_iget fails, then goto next to find inode(DF). 8. CP | dnode(F) | inode(x) -> If f2fs_iget fails, then goto next to find inode(DF). But it will fail due to no inode(DF). So, this patch adds some missing points such as #1, #5, #7, and #8. Signed-off-by: Huang Ying <ying.huang@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-16 07:46:08 +08:00
struct f2fs_inode *raw = F2FS_INODE(page);
inode->i_mode = le16_to_cpu(raw->i_mode);
i_size_write(inode, le64_to_cpu(raw->i_size));
inode->i_atime.tv_sec = le64_to_cpu(raw->i_mtime);
inode->i_ctime.tv_sec = le64_to_cpu(raw->i_ctime);
inode->i_mtime.tv_sec = le64_to_cpu(raw->i_mtime);
inode->i_atime.tv_nsec = le32_to_cpu(raw->i_mtime_nsec);
inode->i_ctime.tv_nsec = le32_to_cpu(raw->i_ctime_nsec);
inode->i_mtime.tv_nsec = le32_to_cpu(raw->i_mtime_nsec);
f2fs: add debug msgs in the recovery routine This patch adds some trivial debugging messages in the recovery process. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-16 14:04:49 +08:00
f2fs_msg(inode->i_sb, KERN_NOTICE, "recover_inode: ino = %x, name = %s",
f2fs: remove redundant operation during roll-forward recovery If same data is updated multiple times, we don't need to redo whole the operations. Let's just update the lastest one. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 05:29:06 +08:00
ino_of_node(page), F2FS_INODE(page)->i_name);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
}
static int find_fsync_dnodes(struct f2fs_sb_info *sbi, struct list_head *head)
{
f2fs: introduce cur_cp_version function to reduce code size This patch introduces a new inline function, cur_cp_version, to reduce redundant codes. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-08-09 14:03:21 +08:00
unsigned long long cp_ver = cur_cp_version(F2FS_CKPT(sbi));
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
struct curseg_info *curseg;
f2fs: use meta_inode cache to improve roll-forward speed Previously, all the dnode pages should be read during the roll-forward recovery. Even worsely, whole the chain was traversed twice. This patch removes that redundant and costly read operations by using page cache of meta_inode and readahead function as well. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 04:49:55 +08:00
struct page *page = NULL;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
block_t blkaddr;
int err = 0;
/* get node pages in the current segment */
curseg = CURSEG_I(sbi, CURSEG_WARM_NODE);
f2fs: use existing macro to clean up some codes This patch use existing macro F2FS_INODE/NEXT_FREE_BLKADDR to clean up some codes. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-02-27 19:52:21 +08:00
blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
while (1) {
struct fsync_inode_entry *entry;
f2fs: check the use of macros on block counts and addresses This patch cleans up the existing and new macros for readability. Rule is like this. ,-----------------------------------------> MAX_BLKADDR -, | ,------------- TOTAL_BLKS ----------------------------, | | | | ,- seg0_blkaddr ,----- sit/nat/ssa/main blkaddress | block | | (SEG0_BLKADDR) | | | | (e.g., MAIN_BLKADDR) | address 0..x................ a b c d ............................. | | global seg# 0...................... m ............................. | | | | `------- MAIN_SEGS -----------' `-------------- TOTAL_SEGS ---------------------------' | | seg# 0..........xx.................. = Note = o GET_SEGNO_FROM_SEG0 : blk address -> global segno o GET_SEGNO : blk address -> segno o START_BLOCK : segno -> starting block address Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-24 02:23:01 +08:00
if (blkaddr < MAIN_BLKADDR(sbi) || blkaddr >= MAX_BLKADDR(sbi))
f2fs: use meta_inode cache to improve roll-forward speed Previously, all the dnode pages should be read during the roll-forward recovery. Even worsely, whole the chain was traversed twice. This patch removes that redundant and costly read operations by using page cache of meta_inode and readahead function as well. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 04:49:55 +08:00
return 0;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
f2fs: use meta_inode cache to improve roll-forward speed Previously, all the dnode pages should be read during the roll-forward recovery. Even worsely, whole the chain was traversed twice. This patch removes that redundant and costly read operations by using page cache of meta_inode and readahead function as well. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 04:49:55 +08:00
page = get_meta_page_ra(sbi, blkaddr);
f2fs: reduce unncessary locking pages during read This patch reduces redundant locking and unlocking pages during read operations. In f2fs_readpage, let's use wait_on_page_locked() instead of lock_page. And then, when we need to modify any data finally, let's lock the page so that we can avoid lock contention. [readpage rule] - The f2fs_readpage returns unlocked page, or released page too in error cases. - Its caller should handle read error, -EIO, after locking the page, which indicates read completion. - Its caller should check PageUptodate after grab_cache_page. Signed-off-by: Changman Lee <cm224.lee@samsung.com> Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-08 20:29:23 +08:00
f2fs: fix the recovery flow to handle errors correctly We should handle errors during the recovery flow correctly. For example, if we get -ENOMEM, we should report a mount failure instead of conducting the remained mount procedure. Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-20 18:01:06 +08:00
if (cp_ver != cpver_of_node(page))
f2fs: add debug msgs in the recovery routine This patch adds some trivial debugging messages in the recovery process. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-16 14:04:49 +08:00
break;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
if (!is_fsync_dnode(page))
goto next;
entry = get_fsync_inode(head, ino_of_node(page));
if (entry) {
if (IS_INODE(page) && is_dent_dnode(page))
set_inode_flag(F2FS_I(entry->inode),
FI_INC_LINK);
} else {
if (IS_INODE(page) && is_dent_dnode(page)) {
f2fs: fix the recovery flow to handle errors correctly We should handle errors during the recovery flow correctly. For example, if we get -ENOMEM, we should report a mount failure instead of conducting the remained mount procedure. Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-20 18:01:06 +08:00
err = recover_inode_page(sbi, page);
if (err)
f2fs: add debug msgs in the recovery routine This patch adds some trivial debugging messages in the recovery process. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-16 14:04:49 +08:00
break;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
}
/* add this fsync inode to the list */
f2fs: remove redundant operation during roll-forward recovery If same data is updated multiple times, we don't need to redo whole the operations. Let's just update the lastest one. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 05:29:06 +08:00
entry = kmem_cache_alloc(fsync_entry_slab, GFP_F2FS_ZERO);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
if (!entry) {
err = -ENOMEM;
f2fs: add debug msgs in the recovery routine This patch adds some trivial debugging messages in the recovery process. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-16 14:04:49 +08:00
break;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
}
f2fs: fix roll-forward missing scenarios We can summarize the roll forward recovery scenarios as follows. [Term] F: fsync_mark, D: dentry_mark 1. inode(x) | CP | inode(x) | dnode(F) -> Update the latest inode(x). 2. inode(x) | CP | inode(F) | dnode(F) -> No problem. 3. inode(x) | CP | dnode(F) | inode(x) -> Recover to the latest dnode(F), and drop the last inode(x) 4. inode(x) | CP | dnode(F) | inode(F) -> No problem. 5. CP | inode(x) | dnode(F) -> The inode(DF) was missing. Should drop this dnode(F). 6. CP | inode(DF) | dnode(F) -> No problem. 7. CP | dnode(F) | inode(DF) -> If f2fs_iget fails, then goto next to find inode(DF). 8. CP | dnode(F) | inode(x) -> If f2fs_iget fails, then goto next to find inode(DF). But it will fail due to no inode(DF). So, this patch adds some missing points such as #1, #5, #7, and #8. Signed-off-by: Huang Ying <ying.huang@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-16 07:46:08 +08:00
/*
* CP | dnode(F) | inode(DF)
* For this case, we should not give up now.
*/
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
entry->inode = f2fs_iget(sbi->sb, ino_of_node(page));
if (IS_ERR(entry->inode)) {
err = PTR_ERR(entry->inode);
f2fs: fix fsync_inode list addition logic and avoid invalid access to memory In function find_fsync_dnodes() - the fsync inodes gets added to the list, but in one path suppose f2fs_iget results in error, in such case - error gets added to the fsync inode list. In next call to recover_data()->get_fsync_inode() entry = list_entry(this, struct fsync_inode_entry, list); if (entry->inode->i_ino == ino) This can result in "invalid access to memory" when it encounters 'error' as entry in the fsync inode list. So, add the fsync inode entry to the list only in case of no errors. And, free the object at that point itself in case of issue. Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Amit Sahrawat <a.sahrawat@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-12-22 11:10:12 +08:00
kmem_cache_free(fsync_entry_slab, entry);
f2fs: fix roll-forward missing scenarios We can summarize the roll forward recovery scenarios as follows. [Term] F: fsync_mark, D: dentry_mark 1. inode(x) | CP | inode(x) | dnode(F) -> Update the latest inode(x). 2. inode(x) | CP | inode(F) | dnode(F) -> No problem. 3. inode(x) | CP | dnode(F) | inode(x) -> Recover to the latest dnode(F), and drop the last inode(x) 4. inode(x) | CP | dnode(F) | inode(F) -> No problem. 5. CP | inode(x) | dnode(F) -> The inode(DF) was missing. Should drop this dnode(F). 6. CP | inode(DF) | dnode(F) -> No problem. 7. CP | dnode(F) | inode(DF) -> If f2fs_iget fails, then goto next to find inode(DF). 8. CP | dnode(F) | inode(x) -> If f2fs_iget fails, then goto next to find inode(DF). But it will fail due to no inode(DF). So, this patch adds some missing points such as #1, #5, #7, and #8. Signed-off-by: Huang Ying <ying.huang@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-16 07:46:08 +08:00
if (err == -ENOENT)
goto next;
f2fs: add debug msgs in the recovery routine This patch adds some trivial debugging messages in the recovery process. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-16 14:04:49 +08:00
break;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
}
f2fs: fix fsync_inode list addition logic and avoid invalid access to memory In function find_fsync_dnodes() - the fsync inodes gets added to the list, but in one path suppose f2fs_iget results in error, in such case - error gets added to the fsync inode list. In next call to recover_data()->get_fsync_inode() entry = list_entry(this, struct fsync_inode_entry, list); if (entry->inode->i_ino == ino) This can result in "invalid access to memory" when it encounters 'error' as entry in the fsync inode list. So, add the fsync inode entry to the list only in case of no errors. And, free the object at that point itself in case of issue. Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Amit Sahrawat <a.sahrawat@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-12-22 11:10:12 +08:00
list_add_tail(&entry->list, head);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
}
f2fs: remove redundant assignment We don't need to assign a value redundantly. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-15 09:49:13 +08:00
entry->blkaddr = blkaddr;
f2fs: remove redundant operation during roll-forward recovery If same data is updated multiple times, we don't need to redo whole the operations. Let's just update the lastest one. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 05:29:06 +08:00
if (IS_INODE(page)) {
entry->last_inode = blkaddr;
if (is_dent_dnode(page))
entry->last_dentry = blkaddr;
}
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
next:
/* check next segment */
blkaddr = next_blkaddr_of_node(page);
f2fs: use meta_inode cache to improve roll-forward speed Previously, all the dnode pages should be read during the roll-forward recovery. Even worsely, whole the chain was traversed twice. This patch removes that redundant and costly read operations by using page cache of meta_inode and readahead function as well. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 04:49:55 +08:00
f2fs_put_page(page, 1);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
}
f2fs: use meta_inode cache to improve roll-forward speed Previously, all the dnode pages should be read during the roll-forward recovery. Even worsely, whole the chain was traversed twice. This patch removes that redundant and costly read operations by using page cache of meta_inode and readahead function as well. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 04:49:55 +08:00
f2fs_put_page(page, 1);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
return err;
}
f2fs: remove the unused argument "sbi" of func destroy_fsync_dnodes() As destroy_fsync_dnodes() is a simple list-cleanup func, so delete the unused and unrelated f2fs_sb_info argument of it. Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-06-27 09:28:54 +08:00
static void destroy_fsync_dnodes(struct list_head *head)
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
{
f2fs: use _safe() version of list_for_each This is calling list_del() inside a loop which is a problem when we try move to the next item on the list. I've converted it to use the _safe version. And also, as a cleanup, I've converted it to use list_for_each_entry instead of list_for_each. Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Dmitry Torokhov <dmitry.torokhov@gmail.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-01-20 23:02:58 +08:00
struct fsync_inode_entry *entry, *tmp;
list_for_each_entry_safe(entry, tmp, head, list) {
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
iput(entry->inode);
list_del(&entry->list);
kmem_cache_free(fsync_entry_slab, entry);
}
}
f2fs: fix to handle do_recover_data errors This patch adds error handling codes of check_index_in_prev_nodes and its caller, do_recover_data. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:20:01 +08:00
static int check_index_in_prev_nodes(struct f2fs_sb_info *sbi,
f2fs: reuse the locked dnode page and its inode This patch fixes the following deadlock bug during the recovery. INFO: task mount:1322 blocked for more than 120 seconds. "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. mount D ffffffff81125870 0 1322 1266 0x00000000 ffff8801207e39d8 0000000000000046 ffff88012ab1dee0 0000000000000046 ffff8801207e3a08 ffff880115903f40 ffff8801207e3fd8 ffff8801207e3fd8 ffff8801207e3fd8 ffff880115903f40 ffff8801207e39d8 ffff88012fc94520 Call Trace: [<ffffffff81125870>] ? __lock_page+0x70/0x70 [<ffffffff816a92d9>] schedule+0x29/0x70 [<ffffffff816a93af>] io_schedule+0x8f/0xd0 [<ffffffff8112587e>] sleep_on_page+0xe/0x20 [<ffffffff816a649a>] __wait_on_bit_lock+0x5a/0xc0 [<ffffffff81125867>] __lock_page+0x67/0x70 [<ffffffff8106c7b0>] ? autoremove_wake_function+0x40/0x40 [<ffffffff81126857>] find_lock_page+0x67/0x80 [<ffffffff8112698f>] find_or_create_page+0x3f/0xb0 [<ffffffffa03901a8>] ? sync_inode_page+0xa8/0xd0 [f2fs] [<ffffffffa038fdf7>] get_node_page+0x67/0x180 [f2fs] [<ffffffffa039818b>] recover_fsync_data+0xacb/0xff0 [f2fs] [<ffffffff816aaa1e>] ? _raw_spin_unlock+0x3e/0x40 [<ffffffffa0389634>] f2fs_fill_super+0x7d4/0x850 [f2fs] [<ffffffff81184cf9>] mount_bdev+0x1c9/0x210 [<ffffffffa0388e60>] ? validate_superblock+0x180/0x180 [f2fs] [<ffffffffa0387635>] f2fs_mount+0x15/0x20 [f2fs] [<ffffffff81185a13>] mount_fs+0x43/0x1b0 [<ffffffff81145ba0>] ? __alloc_percpu+0x10/0x20 [<ffffffff811a0796>] vfs_kern_mount+0x76/0x120 [<ffffffff811a2cb7>] do_mount+0x237/0xa10 [<ffffffff81140b9b>] ? strndup_user+0x5b/0x80 [<ffffffff811a3520>] SyS_mount+0x90/0xe0 [<ffffffff816b3502>] system_call_fastpath+0x16/0x1b The bug is triggered when check_index_in_prev_nodes tries to get the direct node page by calling get_node_page. At this point, if the direct node page is already locked by get_dnode_of_data, its caller, we got a deadlock condition. This patch adds additional condition check for the reuse of locked direct node pages prior to the get_node_page call. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:02:02 +08:00
block_t blkaddr, struct dnode_of_data *dn)
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
{
struct seg_entry *sentry;
unsigned int segno = GET_SEGNO(sbi, blkaddr);
f2fs: clean up with a macro This patch adds GET_BLKOFF_FROM_SEG0 to clean up some codes. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-02-04 12:01:10 +08:00
unsigned short blkoff = GET_BLKOFF_FROM_SEG0(sbi, blkaddr);
f2fs: fix a build warning This patch modifies flow a little bit to avoid the following build warnings. src/fs/f2fs/recovery.c: In function ‘check_index_in_prev_nodes’: src/fs/f2fs/recovery.c:288:51: warning: ‘sum.<U5390>.<U52f8>.ofs_in_node’ may be used uninitialized in this function [-Wmaybe-uninitialized] src/fs/f2fs/recovery.c:260:23: warning: ‘sum.nid’ may be used uninitialized in this function [-Wmaybe-uninitialized] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-01-28 13:54:07 +08:00
struct f2fs_summary_block *sum_node;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
struct f2fs_summary sum;
f2fs: fix a build warning This patch modifies flow a little bit to avoid the following build warnings. src/fs/f2fs/recovery.c: In function ‘check_index_in_prev_nodes’: src/fs/f2fs/recovery.c:288:51: warning: ‘sum.<U5390>.<U52f8>.ofs_in_node’ may be used uninitialized in this function [-Wmaybe-uninitialized] src/fs/f2fs/recovery.c:260:23: warning: ‘sum.nid’ may be used uninitialized in this function [-Wmaybe-uninitialized] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-01-28 13:54:07 +08:00
struct page *sum_page, *node_page;
f2fs: reuse the locked dnode page and its inode This patch fixes the following deadlock bug during the recovery. INFO: task mount:1322 blocked for more than 120 seconds. "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. mount D ffffffff81125870 0 1322 1266 0x00000000 ffff8801207e39d8 0000000000000046 ffff88012ab1dee0 0000000000000046 ffff8801207e3a08 ffff880115903f40 ffff8801207e3fd8 ffff8801207e3fd8 ffff8801207e3fd8 ffff880115903f40 ffff8801207e39d8 ffff88012fc94520 Call Trace: [<ffffffff81125870>] ? __lock_page+0x70/0x70 [<ffffffff816a92d9>] schedule+0x29/0x70 [<ffffffff816a93af>] io_schedule+0x8f/0xd0 [<ffffffff8112587e>] sleep_on_page+0xe/0x20 [<ffffffff816a649a>] __wait_on_bit_lock+0x5a/0xc0 [<ffffffff81125867>] __lock_page+0x67/0x70 [<ffffffff8106c7b0>] ? autoremove_wake_function+0x40/0x40 [<ffffffff81126857>] find_lock_page+0x67/0x80 [<ffffffff8112698f>] find_or_create_page+0x3f/0xb0 [<ffffffffa03901a8>] ? sync_inode_page+0xa8/0xd0 [f2fs] [<ffffffffa038fdf7>] get_node_page+0x67/0x180 [f2fs] [<ffffffffa039818b>] recover_fsync_data+0xacb/0xff0 [f2fs] [<ffffffff816aaa1e>] ? _raw_spin_unlock+0x3e/0x40 [<ffffffffa0389634>] f2fs_fill_super+0x7d4/0x850 [f2fs] [<ffffffff81184cf9>] mount_bdev+0x1c9/0x210 [<ffffffffa0388e60>] ? validate_superblock+0x180/0x180 [f2fs] [<ffffffffa0387635>] f2fs_mount+0x15/0x20 [f2fs] [<ffffffff81185a13>] mount_fs+0x43/0x1b0 [<ffffffff81145ba0>] ? __alloc_percpu+0x10/0x20 [<ffffffff811a0796>] vfs_kern_mount+0x76/0x120 [<ffffffff811a2cb7>] do_mount+0x237/0xa10 [<ffffffff81140b9b>] ? strndup_user+0x5b/0x80 [<ffffffff811a3520>] SyS_mount+0x90/0xe0 [<ffffffff816b3502>] system_call_fastpath+0x16/0x1b The bug is triggered when check_index_in_prev_nodes tries to get the direct node page by calling get_node_page. At this point, if the direct node page is already locked by get_dnode_of_data, its caller, we got a deadlock condition. This patch adds additional condition check for the reuse of locked direct node pages prior to the get_node_page call. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:02:02 +08:00
nid_t ino, nid;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
struct inode *inode;
f2fs: reserve the xattr space dynamically This patch enables the number of direct pointers inside on-disk inode block to be changed dynamically according to the size of inline xattr space. The number of direct pointers, ADDRS_PER_INODE, can be changed only if the file has inline xattr flag. The number of direct pointers that will be used by inline xattrs is defined as F2FS_INLINE_XATTR_ADDRS. Current patch assigns F2FS_INLINE_XATTR_ADDRS to 0 temporarily. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-08-12 20:08:03 +08:00
unsigned int offset;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
block_t bidx;
int i;
sentry = get_seg_entry(sbi, segno);
if (!f2fs_test_bit(blkoff, sentry->cur_valid_map))
f2fs: fix to handle do_recover_data errors This patch adds error handling codes of check_index_in_prev_nodes and its caller, do_recover_data. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:20:01 +08:00
return 0;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
/* Get the previous summary */
for (i = CURSEG_WARM_DATA; i <= CURSEG_COLD_DATA; i++) {
struct curseg_info *curseg = CURSEG_I(sbi, i);
if (curseg->segno == segno) {
sum = curseg->sum_blk->entries[blkoff];
f2fs: fix a build warning This patch modifies flow a little bit to avoid the following build warnings. src/fs/f2fs/recovery.c: In function ‘check_index_in_prev_nodes’: src/fs/f2fs/recovery.c:288:51: warning: ‘sum.<U5390>.<U52f8>.ofs_in_node’ may be used uninitialized in this function [-Wmaybe-uninitialized] src/fs/f2fs/recovery.c:260:23: warning: ‘sum.nid’ may be used uninitialized in this function [-Wmaybe-uninitialized] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-01-28 13:54:07 +08:00
goto got_it;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
}
}
f2fs: fix a build warning This patch modifies flow a little bit to avoid the following build warnings. src/fs/f2fs/recovery.c: In function ‘check_index_in_prev_nodes’: src/fs/f2fs/recovery.c:288:51: warning: ‘sum.<U5390>.<U52f8>.ofs_in_node’ may be used uninitialized in this function [-Wmaybe-uninitialized] src/fs/f2fs/recovery.c:260:23: warning: ‘sum.nid’ may be used uninitialized in this function [-Wmaybe-uninitialized] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-01-28 13:54:07 +08:00
sum_page = get_sum_page(sbi, segno);
sum_node = (struct f2fs_summary_block *)page_address(sum_page);
sum = sum_node->entries[blkoff];
f2fs_put_page(sum_page, 1);
got_it:
f2fs: reuse the locked dnode page and its inode This patch fixes the following deadlock bug during the recovery. INFO: task mount:1322 blocked for more than 120 seconds. "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. mount D ffffffff81125870 0 1322 1266 0x00000000 ffff8801207e39d8 0000000000000046 ffff88012ab1dee0 0000000000000046 ffff8801207e3a08 ffff880115903f40 ffff8801207e3fd8 ffff8801207e3fd8 ffff8801207e3fd8 ffff880115903f40 ffff8801207e39d8 ffff88012fc94520 Call Trace: [<ffffffff81125870>] ? __lock_page+0x70/0x70 [<ffffffff816a92d9>] schedule+0x29/0x70 [<ffffffff816a93af>] io_schedule+0x8f/0xd0 [<ffffffff8112587e>] sleep_on_page+0xe/0x20 [<ffffffff816a649a>] __wait_on_bit_lock+0x5a/0xc0 [<ffffffff81125867>] __lock_page+0x67/0x70 [<ffffffff8106c7b0>] ? autoremove_wake_function+0x40/0x40 [<ffffffff81126857>] find_lock_page+0x67/0x80 [<ffffffff8112698f>] find_or_create_page+0x3f/0xb0 [<ffffffffa03901a8>] ? sync_inode_page+0xa8/0xd0 [f2fs] [<ffffffffa038fdf7>] get_node_page+0x67/0x180 [f2fs] [<ffffffffa039818b>] recover_fsync_data+0xacb/0xff0 [f2fs] [<ffffffff816aaa1e>] ? _raw_spin_unlock+0x3e/0x40 [<ffffffffa0389634>] f2fs_fill_super+0x7d4/0x850 [f2fs] [<ffffffff81184cf9>] mount_bdev+0x1c9/0x210 [<ffffffffa0388e60>] ? validate_superblock+0x180/0x180 [f2fs] [<ffffffffa0387635>] f2fs_mount+0x15/0x20 [f2fs] [<ffffffff81185a13>] mount_fs+0x43/0x1b0 [<ffffffff81145ba0>] ? __alloc_percpu+0x10/0x20 [<ffffffff811a0796>] vfs_kern_mount+0x76/0x120 [<ffffffff811a2cb7>] do_mount+0x237/0xa10 [<ffffffff81140b9b>] ? strndup_user+0x5b/0x80 [<ffffffff811a3520>] SyS_mount+0x90/0xe0 [<ffffffff816b3502>] system_call_fastpath+0x16/0x1b The bug is triggered when check_index_in_prev_nodes tries to get the direct node page by calling get_node_page. At this point, if the direct node page is already locked by get_dnode_of_data, its caller, we got a deadlock condition. This patch adds additional condition check for the reuse of locked direct node pages prior to the get_node_page call. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:02:02 +08:00
/* Use the locked dnode page and inode */
nid = le32_to_cpu(sum.nid);
if (dn->inode->i_ino == nid) {
struct dnode_of_data tdn = *dn;
tdn.nid = nid;
tdn.node_page = dn->inode_page;
f2fs: fix an endian conversion bug detected by sparse This patch should fix the following bug reported by kbuild test robot. fs/f2fs/recovery.c:233:33: sparse: incorrect type in assignment (different base types) parse warnings: (new ones prefixed by >>) >> recovery.c:233: sparse: incorrect type in assignment (different base types) recovery.c:233: expected unsigned int [unsigned] [assigned] ofs_in_node recovery.c:233: got restricted __le16 [assigned] [usertype] ofs_in_node >> recovery.c:238: sparse: incorrect type in assignment (different base types) recovery.c:238: expected unsigned int [unsigned] ofs_in_node recovery.c:238: got restricted __le16 [assigned] [usertype] ofs_in_node Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-06-24 06:47:23 +08:00
tdn.ofs_in_node = le16_to_cpu(sum.ofs_in_node);
f2fs: reuse the locked dnode page and its inode This patch fixes the following deadlock bug during the recovery. INFO: task mount:1322 blocked for more than 120 seconds. "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. mount D ffffffff81125870 0 1322 1266 0x00000000 ffff8801207e39d8 0000000000000046 ffff88012ab1dee0 0000000000000046 ffff8801207e3a08 ffff880115903f40 ffff8801207e3fd8 ffff8801207e3fd8 ffff8801207e3fd8 ffff880115903f40 ffff8801207e39d8 ffff88012fc94520 Call Trace: [<ffffffff81125870>] ? __lock_page+0x70/0x70 [<ffffffff816a92d9>] schedule+0x29/0x70 [<ffffffff816a93af>] io_schedule+0x8f/0xd0 [<ffffffff8112587e>] sleep_on_page+0xe/0x20 [<ffffffff816a649a>] __wait_on_bit_lock+0x5a/0xc0 [<ffffffff81125867>] __lock_page+0x67/0x70 [<ffffffff8106c7b0>] ? autoremove_wake_function+0x40/0x40 [<ffffffff81126857>] find_lock_page+0x67/0x80 [<ffffffff8112698f>] find_or_create_page+0x3f/0xb0 [<ffffffffa03901a8>] ? sync_inode_page+0xa8/0xd0 [f2fs] [<ffffffffa038fdf7>] get_node_page+0x67/0x180 [f2fs] [<ffffffffa039818b>] recover_fsync_data+0xacb/0xff0 [f2fs] [<ffffffff816aaa1e>] ? _raw_spin_unlock+0x3e/0x40 [<ffffffffa0389634>] f2fs_fill_super+0x7d4/0x850 [f2fs] [<ffffffff81184cf9>] mount_bdev+0x1c9/0x210 [<ffffffffa0388e60>] ? validate_superblock+0x180/0x180 [f2fs] [<ffffffffa0387635>] f2fs_mount+0x15/0x20 [f2fs] [<ffffffff81185a13>] mount_fs+0x43/0x1b0 [<ffffffff81145ba0>] ? __alloc_percpu+0x10/0x20 [<ffffffff811a0796>] vfs_kern_mount+0x76/0x120 [<ffffffff811a2cb7>] do_mount+0x237/0xa10 [<ffffffff81140b9b>] ? strndup_user+0x5b/0x80 [<ffffffff811a3520>] SyS_mount+0x90/0xe0 [<ffffffff816b3502>] system_call_fastpath+0x16/0x1b The bug is triggered when check_index_in_prev_nodes tries to get the direct node page by calling get_node_page. At this point, if the direct node page is already locked by get_dnode_of_data, its caller, we got a deadlock condition. This patch adds additional condition check for the reuse of locked direct node pages prior to the get_node_page call. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:02:02 +08:00
truncate_data_blocks_range(&tdn, 1);
f2fs: fix to handle do_recover_data errors This patch adds error handling codes of check_index_in_prev_nodes and its caller, do_recover_data. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:20:01 +08:00
return 0;
f2fs: reuse the locked dnode page and its inode This patch fixes the following deadlock bug during the recovery. INFO: task mount:1322 blocked for more than 120 seconds. "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. mount D ffffffff81125870 0 1322 1266 0x00000000 ffff8801207e39d8 0000000000000046 ffff88012ab1dee0 0000000000000046 ffff8801207e3a08 ffff880115903f40 ffff8801207e3fd8 ffff8801207e3fd8 ffff8801207e3fd8 ffff880115903f40 ffff8801207e39d8 ffff88012fc94520 Call Trace: [<ffffffff81125870>] ? __lock_page+0x70/0x70 [<ffffffff816a92d9>] schedule+0x29/0x70 [<ffffffff816a93af>] io_schedule+0x8f/0xd0 [<ffffffff8112587e>] sleep_on_page+0xe/0x20 [<ffffffff816a649a>] __wait_on_bit_lock+0x5a/0xc0 [<ffffffff81125867>] __lock_page+0x67/0x70 [<ffffffff8106c7b0>] ? autoremove_wake_function+0x40/0x40 [<ffffffff81126857>] find_lock_page+0x67/0x80 [<ffffffff8112698f>] find_or_create_page+0x3f/0xb0 [<ffffffffa03901a8>] ? sync_inode_page+0xa8/0xd0 [f2fs] [<ffffffffa038fdf7>] get_node_page+0x67/0x180 [f2fs] [<ffffffffa039818b>] recover_fsync_data+0xacb/0xff0 [f2fs] [<ffffffff816aaa1e>] ? _raw_spin_unlock+0x3e/0x40 [<ffffffffa0389634>] f2fs_fill_super+0x7d4/0x850 [f2fs] [<ffffffff81184cf9>] mount_bdev+0x1c9/0x210 [<ffffffffa0388e60>] ? validate_superblock+0x180/0x180 [f2fs] [<ffffffffa0387635>] f2fs_mount+0x15/0x20 [f2fs] [<ffffffff81185a13>] mount_fs+0x43/0x1b0 [<ffffffff81145ba0>] ? __alloc_percpu+0x10/0x20 [<ffffffff811a0796>] vfs_kern_mount+0x76/0x120 [<ffffffff811a2cb7>] do_mount+0x237/0xa10 [<ffffffff81140b9b>] ? strndup_user+0x5b/0x80 [<ffffffff811a3520>] SyS_mount+0x90/0xe0 [<ffffffff816b3502>] system_call_fastpath+0x16/0x1b The bug is triggered when check_index_in_prev_nodes tries to get the direct node page by calling get_node_page. At this point, if the direct node page is already locked by get_dnode_of_data, its caller, we got a deadlock condition. This patch adds additional condition check for the reuse of locked direct node pages prior to the get_node_page call. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:02:02 +08:00
} else if (dn->nid == nid) {
struct dnode_of_data tdn = *dn;
f2fs: fix an endian conversion bug detected by sparse This patch should fix the following bug reported by kbuild test robot. fs/f2fs/recovery.c:233:33: sparse: incorrect type in assignment (different base types) parse warnings: (new ones prefixed by >>) >> recovery.c:233: sparse: incorrect type in assignment (different base types) recovery.c:233: expected unsigned int [unsigned] [assigned] ofs_in_node recovery.c:233: got restricted __le16 [assigned] [usertype] ofs_in_node >> recovery.c:238: sparse: incorrect type in assignment (different base types) recovery.c:238: expected unsigned int [unsigned] ofs_in_node recovery.c:238: got restricted __le16 [assigned] [usertype] ofs_in_node Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-06-24 06:47:23 +08:00
tdn.ofs_in_node = le16_to_cpu(sum.ofs_in_node);
f2fs: reuse the locked dnode page and its inode This patch fixes the following deadlock bug during the recovery. INFO: task mount:1322 blocked for more than 120 seconds. "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. mount D ffffffff81125870 0 1322 1266 0x00000000 ffff8801207e39d8 0000000000000046 ffff88012ab1dee0 0000000000000046 ffff8801207e3a08 ffff880115903f40 ffff8801207e3fd8 ffff8801207e3fd8 ffff8801207e3fd8 ffff880115903f40 ffff8801207e39d8 ffff88012fc94520 Call Trace: [<ffffffff81125870>] ? __lock_page+0x70/0x70 [<ffffffff816a92d9>] schedule+0x29/0x70 [<ffffffff816a93af>] io_schedule+0x8f/0xd0 [<ffffffff8112587e>] sleep_on_page+0xe/0x20 [<ffffffff816a649a>] __wait_on_bit_lock+0x5a/0xc0 [<ffffffff81125867>] __lock_page+0x67/0x70 [<ffffffff8106c7b0>] ? autoremove_wake_function+0x40/0x40 [<ffffffff81126857>] find_lock_page+0x67/0x80 [<ffffffff8112698f>] find_or_create_page+0x3f/0xb0 [<ffffffffa03901a8>] ? sync_inode_page+0xa8/0xd0 [f2fs] [<ffffffffa038fdf7>] get_node_page+0x67/0x180 [f2fs] [<ffffffffa039818b>] recover_fsync_data+0xacb/0xff0 [f2fs] [<ffffffff816aaa1e>] ? _raw_spin_unlock+0x3e/0x40 [<ffffffffa0389634>] f2fs_fill_super+0x7d4/0x850 [f2fs] [<ffffffff81184cf9>] mount_bdev+0x1c9/0x210 [<ffffffffa0388e60>] ? validate_superblock+0x180/0x180 [f2fs] [<ffffffffa0387635>] f2fs_mount+0x15/0x20 [f2fs] [<ffffffff81185a13>] mount_fs+0x43/0x1b0 [<ffffffff81145ba0>] ? __alloc_percpu+0x10/0x20 [<ffffffff811a0796>] vfs_kern_mount+0x76/0x120 [<ffffffff811a2cb7>] do_mount+0x237/0xa10 [<ffffffff81140b9b>] ? strndup_user+0x5b/0x80 [<ffffffff811a3520>] SyS_mount+0x90/0xe0 [<ffffffff816b3502>] system_call_fastpath+0x16/0x1b The bug is triggered when check_index_in_prev_nodes tries to get the direct node page by calling get_node_page. At this point, if the direct node page is already locked by get_dnode_of_data, its caller, we got a deadlock condition. This patch adds additional condition check for the reuse of locked direct node pages prior to the get_node_page call. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:02:02 +08:00
truncate_data_blocks_range(&tdn, 1);
f2fs: fix to handle do_recover_data errors This patch adds error handling codes of check_index_in_prev_nodes and its caller, do_recover_data. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:20:01 +08:00
return 0;
f2fs: reuse the locked dnode page and its inode This patch fixes the following deadlock bug during the recovery. INFO: task mount:1322 blocked for more than 120 seconds. "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. mount D ffffffff81125870 0 1322 1266 0x00000000 ffff8801207e39d8 0000000000000046 ffff88012ab1dee0 0000000000000046 ffff8801207e3a08 ffff880115903f40 ffff8801207e3fd8 ffff8801207e3fd8 ffff8801207e3fd8 ffff880115903f40 ffff8801207e39d8 ffff88012fc94520 Call Trace: [<ffffffff81125870>] ? __lock_page+0x70/0x70 [<ffffffff816a92d9>] schedule+0x29/0x70 [<ffffffff816a93af>] io_schedule+0x8f/0xd0 [<ffffffff8112587e>] sleep_on_page+0xe/0x20 [<ffffffff816a649a>] __wait_on_bit_lock+0x5a/0xc0 [<ffffffff81125867>] __lock_page+0x67/0x70 [<ffffffff8106c7b0>] ? autoremove_wake_function+0x40/0x40 [<ffffffff81126857>] find_lock_page+0x67/0x80 [<ffffffff8112698f>] find_or_create_page+0x3f/0xb0 [<ffffffffa03901a8>] ? sync_inode_page+0xa8/0xd0 [f2fs] [<ffffffffa038fdf7>] get_node_page+0x67/0x180 [f2fs] [<ffffffffa039818b>] recover_fsync_data+0xacb/0xff0 [f2fs] [<ffffffff816aaa1e>] ? _raw_spin_unlock+0x3e/0x40 [<ffffffffa0389634>] f2fs_fill_super+0x7d4/0x850 [f2fs] [<ffffffff81184cf9>] mount_bdev+0x1c9/0x210 [<ffffffffa0388e60>] ? validate_superblock+0x180/0x180 [f2fs] [<ffffffffa0387635>] f2fs_mount+0x15/0x20 [f2fs] [<ffffffff81185a13>] mount_fs+0x43/0x1b0 [<ffffffff81145ba0>] ? __alloc_percpu+0x10/0x20 [<ffffffff811a0796>] vfs_kern_mount+0x76/0x120 [<ffffffff811a2cb7>] do_mount+0x237/0xa10 [<ffffffff81140b9b>] ? strndup_user+0x5b/0x80 [<ffffffff811a3520>] SyS_mount+0x90/0xe0 [<ffffffff816b3502>] system_call_fastpath+0x16/0x1b The bug is triggered when check_index_in_prev_nodes tries to get the direct node page by calling get_node_page. At this point, if the direct node page is already locked by get_dnode_of_data, its caller, we got a deadlock condition. This patch adds additional condition check for the reuse of locked direct node pages prior to the get_node_page call. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:02:02 +08:00
}
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
/* Get the node page */
f2fs: reuse the locked dnode page and its inode This patch fixes the following deadlock bug during the recovery. INFO: task mount:1322 blocked for more than 120 seconds. "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. mount D ffffffff81125870 0 1322 1266 0x00000000 ffff8801207e39d8 0000000000000046 ffff88012ab1dee0 0000000000000046 ffff8801207e3a08 ffff880115903f40 ffff8801207e3fd8 ffff8801207e3fd8 ffff8801207e3fd8 ffff880115903f40 ffff8801207e39d8 ffff88012fc94520 Call Trace: [<ffffffff81125870>] ? __lock_page+0x70/0x70 [<ffffffff816a92d9>] schedule+0x29/0x70 [<ffffffff816a93af>] io_schedule+0x8f/0xd0 [<ffffffff8112587e>] sleep_on_page+0xe/0x20 [<ffffffff816a649a>] __wait_on_bit_lock+0x5a/0xc0 [<ffffffff81125867>] __lock_page+0x67/0x70 [<ffffffff8106c7b0>] ? autoremove_wake_function+0x40/0x40 [<ffffffff81126857>] find_lock_page+0x67/0x80 [<ffffffff8112698f>] find_or_create_page+0x3f/0xb0 [<ffffffffa03901a8>] ? sync_inode_page+0xa8/0xd0 [f2fs] [<ffffffffa038fdf7>] get_node_page+0x67/0x180 [f2fs] [<ffffffffa039818b>] recover_fsync_data+0xacb/0xff0 [f2fs] [<ffffffff816aaa1e>] ? _raw_spin_unlock+0x3e/0x40 [<ffffffffa0389634>] f2fs_fill_super+0x7d4/0x850 [f2fs] [<ffffffff81184cf9>] mount_bdev+0x1c9/0x210 [<ffffffffa0388e60>] ? validate_superblock+0x180/0x180 [f2fs] [<ffffffffa0387635>] f2fs_mount+0x15/0x20 [f2fs] [<ffffffff81185a13>] mount_fs+0x43/0x1b0 [<ffffffff81145ba0>] ? __alloc_percpu+0x10/0x20 [<ffffffff811a0796>] vfs_kern_mount+0x76/0x120 [<ffffffff811a2cb7>] do_mount+0x237/0xa10 [<ffffffff81140b9b>] ? strndup_user+0x5b/0x80 [<ffffffff811a3520>] SyS_mount+0x90/0xe0 [<ffffffff816b3502>] system_call_fastpath+0x16/0x1b The bug is triggered when check_index_in_prev_nodes tries to get the direct node page by calling get_node_page. At this point, if the direct node page is already locked by get_dnode_of_data, its caller, we got a deadlock condition. This patch adds additional condition check for the reuse of locked direct node pages prior to the get_node_page call. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:02:02 +08:00
node_page = get_node_page(sbi, nid);
f2fs: fix to handle do_recover_data errors This patch adds error handling codes of check_index_in_prev_nodes and its caller, do_recover_data. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:20:01 +08:00
if (IS_ERR(node_page))
return PTR_ERR(node_page);
f2fs: reserve the xattr space dynamically This patch enables the number of direct pointers inside on-disk inode block to be changed dynamically according to the size of inline xattr space. The number of direct pointers, ADDRS_PER_INODE, can be changed only if the file has inline xattr flag. The number of direct pointers that will be used by inline xattrs is defined as F2FS_INLINE_XATTR_ADDRS. Current patch assigns F2FS_INLINE_XATTR_ADDRS to 0 temporarily. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-08-12 20:08:03 +08:00
offset = ofs_of_node(node_page);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
ino = ino_of_node(node_page);
f2fs_put_page(node_page, 1);
f2fs: fix double lock for inode page during roll-foward recovery If the inode is same and its data index are needed to truncate, we can fall into double lock for its inode page via get_dnode_of_data. Error case is like this. 1. write data 1, 2, 3, 4, 5 in inode #4. 2. write data 100, 102, 103, 104, 105 in dnode #6 of inode #4. 3. sync 4. update data 100->106 in dnode #6. 5. fsync inode #4. 6. power-cut -> Then, 1. go back to #3's checkpoint 2. in do_recover_data, get_dnode_of_data() gets inode #4. 3. detect 100->106 in dnode #6. 4. check_index_in_prev_nodes tries to truncate 100 in dnode #6. 5. to trigger truncate_hole, get_dnode_of_data should grab inode #4. 6. detect *kernel hang* This patch should resolve that bug. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 23:35:58 +08:00
if (ino != dn->inode->i_ino) {
/* Deallocate previous index in the node page */
inode = f2fs_iget(sbi->sb, ino);
if (IS_ERR(inode))
return PTR_ERR(inode);
} else {
inode = dn->inode;
}
f2fs: handle error from f2fs_iget_nowait In case f2fs_iget_nowait returns error, it results in truncate_hole being called with 'error' value as inode pointer. There is no check in truncate_hole for valid inode, so it could result in crash due "invalid access to memory". Avoid this by handling error condition properly. Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Amit Sahrawat <a.sahrawat@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-12-22 11:09:43 +08:00
f2fs: reserve the xattr space dynamically This patch enables the number of direct pointers inside on-disk inode block to be changed dynamically according to the size of inline xattr space. The number of direct pointers, ADDRS_PER_INODE, can be changed only if the file has inline xattr flag. The number of direct pointers that will be used by inline xattrs is defined as F2FS_INLINE_XATTR_ADDRS. Current patch assigns F2FS_INLINE_XATTR_ADDRS to 0 temporarily. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-08-12 20:08:03 +08:00
bidx = start_bidx_of_node(offset, F2FS_I(inode)) +
f2fs: fix double lock for inode page during roll-foward recovery If the inode is same and its data index are needed to truncate, we can fall into double lock for its inode page via get_dnode_of_data. Error case is like this. 1. write data 1, 2, 3, 4, 5 in inode #4. 2. write data 100, 102, 103, 104, 105 in dnode #6 of inode #4. 3. sync 4. update data 100->106 in dnode #6. 5. fsync inode #4. 6. power-cut -> Then, 1. go back to #3's checkpoint 2. in do_recover_data, get_dnode_of_data() gets inode #4. 3. detect 100->106 in dnode #6. 4. check_index_in_prev_nodes tries to truncate 100 in dnode #6. 5. to trigger truncate_hole, get_dnode_of_data should grab inode #4. 6. detect *kernel hang* This patch should resolve that bug. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 23:35:58 +08:00
le16_to_cpu(sum.ofs_in_node);
f2fs: reserve the xattr space dynamically This patch enables the number of direct pointers inside on-disk inode block to be changed dynamically according to the size of inline xattr space. The number of direct pointers, ADDRS_PER_INODE, can be changed only if the file has inline xattr flag. The number of direct pointers that will be used by inline xattrs is defined as F2FS_INLINE_XATTR_ADDRS. Current patch assigns F2FS_INLINE_XATTR_ADDRS to 0 temporarily. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-08-12 20:08:03 +08:00
f2fs: fix double lock for inode page during roll-foward recovery If the inode is same and its data index are needed to truncate, we can fall into double lock for its inode page via get_dnode_of_data. Error case is like this. 1. write data 1, 2, 3, 4, 5 in inode #4. 2. write data 100, 102, 103, 104, 105 in dnode #6 of inode #4. 3. sync 4. update data 100->106 in dnode #6. 5. fsync inode #4. 6. power-cut -> Then, 1. go back to #3's checkpoint 2. in do_recover_data, get_dnode_of_data() gets inode #4. 3. detect 100->106 in dnode #6. 4. check_index_in_prev_nodes tries to truncate 100 in dnode #6. 5. to trigger truncate_hole, get_dnode_of_data should grab inode #4. 6. detect *kernel hang* This patch should resolve that bug. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 23:35:58 +08:00
if (ino != dn->inode->i_ino) {
truncate_hole(inode, bidx, bidx + 1);
iput(inode);
} else {
struct dnode_of_data tdn;
set_new_dnode(&tdn, inode, dn->inode_page, NULL, 0);
if (get_dnode_of_data(&tdn, bidx, LOOKUP_NODE))
return 0;
if (tdn.data_blkaddr != NULL_ADDR)
truncate_data_blocks_range(&tdn, 1);
f2fs_put_page(tdn.node_page, 1);
}
f2fs: fix to handle do_recover_data errors This patch adds error handling codes of check_index_in_prev_nodes and its caller, do_recover_data. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:20:01 +08:00
return 0;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
}
f2fs: fix the recovery flow to handle errors correctly We should handle errors during the recovery flow correctly. For example, if we get -ENOMEM, we should report a mount failure instead of conducting the remained mount procedure. Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-20 18:01:06 +08:00
static int do_recover_data(struct f2fs_sb_info *sbi, struct inode *inode,
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
struct page *page, block_t blkaddr)
{
f2fs: reserve the xattr space dynamically This patch enables the number of direct pointers inside on-disk inode block to be changed dynamically according to the size of inline xattr space. The number of direct pointers, ADDRS_PER_INODE, can be changed only if the file has inline xattr flag. The number of direct pointers that will be used by inline xattrs is defined as F2FS_INLINE_XATTR_ADDRS. Current patch assigns F2FS_INLINE_XATTR_ADDRS to 0 temporarily. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-08-12 20:08:03 +08:00
struct f2fs_inode_info *fi = F2FS_I(inode);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
unsigned int start, end;
struct dnode_of_data dn;
struct f2fs_summary sum;
struct node_info ni;
f2fs: add debug msgs in the recovery routine This patch adds some trivial debugging messages in the recovery process. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-16 14:04:49 +08:00
int err = 0, recovered = 0;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
f2fs: fix to recover inline_xattr/data and blocks This patch fixes not to skip xattr recovery and inline xattr/data recovery order. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-08-08 14:49:17 +08:00
/* step 1: recover xattr */
if (IS_INODE(page)) {
recover_inline_xattr(inode, page);
} else if (f2fs_has_xattr_block(ofs_of_node(page))) {
recover_xattr_data(inode, page, blkaddr);
f2fs: add inline_data recovery routine This patch adds a inline_data recovery routine with the following policy. [prev.] [next] of inline_data flag o o -> recover inline_data o x -> remove inline_data, and then recover data blocks x o -> remove inline_data, and then recover inline_data x x -> recover data blocks Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-12-26 11:49:48 +08:00
goto out;
f2fs: fix to recover inline_xattr/data and blocks This patch fixes not to skip xattr recovery and inline xattr/data recovery order. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-08-08 14:49:17 +08:00
}
f2fs: add inline_data recovery routine This patch adds a inline_data recovery routine with the following policy. [prev.] [next] of inline_data flag o o -> recover inline_data o x -> remove inline_data, and then recover data blocks x o -> remove inline_data, and then recover inline_data x x -> recover data blocks Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-12-26 11:49:48 +08:00
f2fs: fix to recover inline_xattr/data and blocks This patch fixes not to skip xattr recovery and inline xattr/data recovery order. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-08-08 14:49:17 +08:00
/* step 2: recover inline data */
if (recover_inline_data(inode, page))
f2fs: fix to recover xattr node block If a new xattr node page was allocated and its inode is fsynced, we should recover the xattr node page during the roll-forward process after power-cut. But, previously, f2fs didn't handle that case, resulting in kernel panic as follows reported by Tom Li. BUG: unable to handle kernel paging request at ffffc9001c861a98 IP: [<ffffffffa0295236>] check_index_in_prev_nodes+0x86/0x2d0 [f2fs] Call Trace: [<ffffffff815ece9b>] ? printk+0x48/0x4a [<ffffffffa029626a>] recover_fsync_data+0xdca/0xf50 [f2fs] [<ffffffffa02873ae>] f2fs_fill_super+0x92e/0x970 [f2fs] [<ffffffff8112c9f8>] mount_bdev+0x1b8/0x200 [<ffffffffa0286a80>] ? f2fs_remount+0x130/0x130 [f2fs] [<ffffffffa0285e40>] f2fs_mount+0x10/0x20 [f2fs] [<ffffffff8112d4de>] mount_fs+0x3e/0x1b0 [<ffffffff810ef4eb>] ? __alloc_percpu+0xb/0x10 [<ffffffff8114761f>] vfs_kern_mount+0x6f/0x120 [<ffffffff811497b9>] do_mount+0x259/0xa90 [<ffffffff810ead1d>] ? memdup_user+0x3d/0x80 [<ffffffff810eadb3>] ? strndup_user+0x53/0x70 [<ffffffff8114a2c9>] SyS_mount+0x89/0xd0 [<ffffffff815feae2>] system_call_fastpath+0x16/0x1b This patch adds a recovery function of xattr node pages. Reported-by: Tom Li <biergaizi@members.fsf.org> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-01-28 11:25:06 +08:00
goto out;
f2fs: fix to recover inline_xattr/data and blocks This patch fixes not to skip xattr recovery and inline xattr/data recovery order. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-08-08 14:49:17 +08:00
/* step 3: recover data indices */
f2fs: reserve the xattr space dynamically This patch enables the number of direct pointers inside on-disk inode block to be changed dynamically according to the size of inline xattr space. The number of direct pointers, ADDRS_PER_INODE, can be changed only if the file has inline xattr flag. The number of direct pointers that will be used by inline xattrs is defined as F2FS_INLINE_XATTR_ADDRS. Current patch assigns F2FS_INLINE_XATTR_ADDRS to 0 temporarily. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-08-12 20:08:03 +08:00
start = start_bidx_of_node(ofs_of_node(page), fi);
f2fs: introduce help macro ADDRS_PER_PAGE() Introduce help macro ADDRS_PER_PAGE() to get the number of address pointers in direct node or inode. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-04-26 19:59:52 +08:00
end = start + ADDRS_PER_PAGE(page, fi);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
f2fs: use rw_sem instead of fs_lock(locks mutex) The fs_locks is used to block other ops(ex, recovery) when doing checkpoint. And each other operate routine(besides checkpoint) needs to acquire a fs_lock, there is a terrible problem here, if these are too many concurrency threads acquiring fs_lock, so that they will block each other and may lead to some performance problem, but this is not the phenomenon we want to see. Though there are some optimization patches introduced to enhance the usage of fs_lock, but the thorough solution is using a *rw_sem* to replace the fs_lock. Checkpoint routine takes write_sem, and other ops take read_sem, so that we can block other ops(ex, recovery) when doing checkpoint, and other ops will not disturb each other, this can avoid the problem described above completely. Because of the weakness of rw_sem, the above change may introduce a potential problem that the checkpoint thread might get starved if other threads are intensively locking the read semaphore for I/O.(Pointed out by Xu Jin) In order to avoid this, a wait_list is introduced, the appending read semaphore ops will be dropped into the wait_list if checkpoint thread is waiting for write semaphore, and will be waked up when checkpoint thread gives up write semaphore. Thanks to Kim's previous review and test, and will be very glad to see other guys' performance tests about this patch. V2: -fix the potential starvation problem. -use more suitable func name suggested by Xu Jin. Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com> [Jaegeuk Kim: adjust minor coding standard] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-09-27 18:08:30 +08:00
f2fs_lock_op(sbi);
f2fs: add inline_data recovery routine This patch adds a inline_data recovery routine with the following policy. [prev.] [next] of inline_data flag o o -> recover inline_data o x -> remove inline_data, and then recover data blocks x o -> remove inline_data, and then recover inline_data x x -> recover data blocks Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-12-26 11:49:48 +08:00
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
set_new_dnode(&dn, inode, NULL, NULL, 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 15:21:29 +08:00
f2fs: fix the recovery flow to handle errors correctly We should handle errors during the recovery flow correctly. For example, if we get -ENOMEM, we should report a mount failure instead of conducting the remained mount procedure. Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-20 18:01:06 +08:00
err = get_dnode_of_data(&dn, start, ALLOC_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 15:21:29 +08:00
if (err) {
f2fs: use rw_sem instead of fs_lock(locks mutex) The fs_locks is used to block other ops(ex, recovery) when doing checkpoint. And each other operate routine(besides checkpoint) needs to acquire a fs_lock, there is a terrible problem here, if these are too many concurrency threads acquiring fs_lock, so that they will block each other and may lead to some performance problem, but this is not the phenomenon we want to see. Though there are some optimization patches introduced to enhance the usage of fs_lock, but the thorough solution is using a *rw_sem* to replace the fs_lock. Checkpoint routine takes write_sem, and other ops take read_sem, so that we can block other ops(ex, recovery) when doing checkpoint, and other ops will not disturb each other, this can avoid the problem described above completely. Because of the weakness of rw_sem, the above change may introduce a potential problem that the checkpoint thread might get starved if other threads are intensively locking the read semaphore for I/O.(Pointed out by Xu Jin) In order to avoid this, a wait_list is introduced, the appending read semaphore ops will be dropped into the wait_list if checkpoint thread is waiting for write semaphore, and will be waked up when checkpoint thread gives up write semaphore. Thanks to Kim's previous review and test, and will be very glad to see other guys' performance tests about this patch. V2: -fix the potential starvation problem. -use more suitable func name suggested by Xu Jin. Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com> [Jaegeuk Kim: adjust minor coding standard] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-09-27 18:08:30 +08:00
f2fs_unlock_op(sbi);
f2fs: add inline_data recovery routine This patch adds a inline_data recovery routine with the following policy. [prev.] [next] of inline_data flag o o -> recover inline_data o x -> remove inline_data, and then recover data blocks x o -> remove inline_data, and then recover inline_data x x -> recover data blocks Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-12-26 11:49:48 +08:00
goto out;
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 15:21:29 +08:00
}
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
f2fs: call f2fs_wait_on_page_writeback instead of native function If a page is on writeback, f2fs can face with deadlock due to under writepages. This is caused by merging IOs inside f2fs, so if it comes to detect, let's throw merged IOs, which is implemented by f2fs_wait_on_page_writeback. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-03-18 12:29:07 +08:00
f2fs_wait_on_page_writeback(dn.node_page, NODE);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
get_node_info(sbi, dn.nid, &ni);
f2fs: need fsck.f2fs when f2fs_bug_on is triggered If any f2fs_bug_on is triggered, fsck.f2fs is needed. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-03 06:52:58 +08:00
f2fs_bug_on(sbi, ni.ino != ino_of_node(page));
f2fs_bug_on(sbi, ofs_of_node(dn.node_page) != ofs_of_node(page));
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
for (; start < end; start++) {
block_t src, dest;
src = datablock_addr(dn.node_page, dn.ofs_in_node);
dest = datablock_addr(page, dn.ofs_in_node);
if (src != dest && dest != NEW_ADDR && dest != NULL_ADDR) {
if (src == NULL_ADDR) {
f2fs: add an option to avoid unnecessary BUG_ONs If you want to remove unnecessary BUG_ONs, you can just turn off F2FS_CHECK_FS in your kernel config. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-29 14:14:54 +08:00
err = reserve_new_block(&dn);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
/* We should not get -ENOSPC */
f2fs: need fsck.f2fs when f2fs_bug_on is triggered If any f2fs_bug_on is triggered, fsck.f2fs is needed. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-03 06:52:58 +08:00
f2fs_bug_on(sbi, err);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
}
/* Check the previous node page having this index */
f2fs: fix to handle do_recover_data errors This patch adds error handling codes of check_index_in_prev_nodes and its caller, do_recover_data. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:20:01 +08:00
err = check_index_in_prev_nodes(sbi, dest, &dn);
if (err)
goto err;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
set_summary(&sum, dn.nid, dn.ofs_in_node, ni.version);
/* write dummy data page */
recover_data_page(sbi, NULL, &sum, src, dest);
update_extent_cache(dest, &dn);
f2fs: add debug msgs in the recovery routine This patch adds some trivial debugging messages in the recovery process. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-16 14:04:49 +08:00
recovered++;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
}
dn.ofs_in_node++;
}
/* write node page in place */
set_summary(&sum, dn.nid, 0, 0);
if (IS_INODE(dn.node_page))
sync_inode_page(&dn);
copy_node_footer(dn.node_page, page);
fill_node_footer(dn.node_page, dn.nid, ni.ino,
ofs_of_node(page), false);
set_page_dirty(dn.node_page);
f2fs: fix to handle do_recover_data errors This patch adds error handling codes of check_index_in_prev_nodes and its caller, do_recover_data. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:20:01 +08:00
err:
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
f2fs_put_dnode(&dn);
f2fs: use rw_sem instead of fs_lock(locks mutex) The fs_locks is used to block other ops(ex, recovery) when doing checkpoint. And each other operate routine(besides checkpoint) needs to acquire a fs_lock, there is a terrible problem here, if these are too many concurrency threads acquiring fs_lock, so that they will block each other and may lead to some performance problem, but this is not the phenomenon we want to see. Though there are some optimization patches introduced to enhance the usage of fs_lock, but the thorough solution is using a *rw_sem* to replace the fs_lock. Checkpoint routine takes write_sem, and other ops take read_sem, so that we can block other ops(ex, recovery) when doing checkpoint, and other ops will not disturb each other, this can avoid the problem described above completely. Because of the weakness of rw_sem, the above change may introduce a potential problem that the checkpoint thread might get starved if other threads are intensively locking the read semaphore for I/O.(Pointed out by Xu Jin) In order to avoid this, a wait_list is introduced, the appending read semaphore ops will be dropped into the wait_list if checkpoint thread is waiting for write semaphore, and will be waked up when checkpoint thread gives up write semaphore. Thanks to Kim's previous review and test, and will be very glad to see other guys' performance tests about this patch. V2: -fix the potential starvation problem. -use more suitable func name suggested by Xu Jin. Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com> [Jaegeuk Kim: adjust minor coding standard] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-09-27 18:08:30 +08:00
f2fs_unlock_op(sbi);
f2fs: add inline_data recovery routine This patch adds a inline_data recovery routine with the following policy. [prev.] [next] of inline_data flag o o -> recover inline_data o x -> remove inline_data, and then recover data blocks x o -> remove inline_data, and then recover inline_data x x -> recover data blocks Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-12-26 11:49:48 +08:00
out:
f2fs: clean checkpatch warnings Fixed a variety of trivial checkpatch warnings. The only delta should be some minor formatting on log strings that were split / too long. Signed-off-by: Chris Fries <cfries@motorola.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-01-18 04:44:39 +08:00
f2fs_msg(sbi->sb, KERN_NOTICE,
"recover_data: ino = %lx, recovered = %d blocks, err = %d",
inode->i_ino, recovered, err);
f2fs: fix to handle do_recover_data errors This patch adds error handling codes of check_index_in_prev_nodes and its caller, do_recover_data. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-22 07:20:01 +08:00
return err;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
}
f2fs: fix the recovery flow to handle errors correctly We should handle errors during the recovery flow correctly. For example, if we get -ENOMEM, we should report a mount failure instead of conducting the remained mount procedure. Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-20 18:01:06 +08:00
static int recover_data(struct f2fs_sb_info *sbi,
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
struct list_head *head, int type)
{
f2fs: introduce cur_cp_version function to reduce code size This patch introduces a new inline function, cur_cp_version, to reduce redundant codes. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-08-09 14:03:21 +08:00
unsigned long long cp_ver = cur_cp_version(F2FS_CKPT(sbi));
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
struct curseg_info *curseg;
f2fs: use meta_inode cache to improve roll-forward speed Previously, all the dnode pages should be read during the roll-forward recovery. Even worsely, whole the chain was traversed twice. This patch removes that redundant and costly read operations by using page cache of meta_inode and readahead function as well. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 04:49:55 +08:00
struct page *page = NULL;
f2fs: fix the recovery flow to handle errors correctly We should handle errors during the recovery flow correctly. For example, if we get -ENOMEM, we should report a mount failure instead of conducting the remained mount procedure. Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-20 18:01:06 +08:00
int err = 0;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
block_t blkaddr;
/* get node pages in the current segment */
curseg = CURSEG_I(sbi, type);
blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
while (1) {
struct fsync_inode_entry *entry;
f2fs: check the use of macros on block counts and addresses This patch cleans up the existing and new macros for readability. Rule is like this. ,-----------------------------------------> MAX_BLKADDR -, | ,------------- TOTAL_BLKS ----------------------------, | | | | ,- seg0_blkaddr ,----- sit/nat/ssa/main blkaddress | block | | (SEG0_BLKADDR) | | | | (e.g., MAIN_BLKADDR) | address 0..x................ a b c d ............................. | | global seg# 0...................... m ............................. | | | | `------- MAIN_SEGS -----------' `-------------- TOTAL_SEGS ---------------------------' | | seg# 0..........xx.................. = Note = o GET_SEGNO_FROM_SEG0 : blk address -> global segno o GET_SEGNO : blk address -> segno o START_BLOCK : segno -> starting block address Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-24 02:23:01 +08:00
if (blkaddr < MAIN_BLKADDR(sbi) || blkaddr >= MAX_BLKADDR(sbi))
f2fs: use meta_inode cache to improve roll-forward speed Previously, all the dnode pages should be read during the roll-forward recovery. Even worsely, whole the chain was traversed twice. This patch removes that redundant and costly read operations by using page cache of meta_inode and readahead function as well. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 04:49:55 +08:00
break;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
f2fs: use meta_inode cache to improve roll-forward speed Previously, all the dnode pages should be read during the roll-forward recovery. Even worsely, whole the chain was traversed twice. This patch removes that redundant and costly read operations by using page cache of meta_inode and readahead function as well. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 04:49:55 +08:00
page = get_meta_page_ra(sbi, blkaddr);
f2fs: reduce unncessary locking pages during read This patch reduces redundant locking and unlocking pages during read operations. In f2fs_readpage, let's use wait_on_page_locked() instead of lock_page. And then, when we need to modify any data finally, let's lock the page so that we can avoid lock contention. [readpage rule] - The f2fs_readpage returns unlocked page, or released page too in error cases. - Its caller should handle read error, -EIO, after locking the page, which indicates read completion. - Its caller should check PageUptodate after grab_cache_page. Signed-off-by: Changman Lee <cm224.lee@samsung.com> Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-08 20:29:23 +08:00
f2fs: use meta_inode cache to improve roll-forward speed Previously, all the dnode pages should be read during the roll-forward recovery. Even worsely, whole the chain was traversed twice. This patch removes that redundant and costly read operations by using page cache of meta_inode and readahead function as well. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 04:49:55 +08:00
if (cp_ver != cpver_of_node(page)) {
f2fs_put_page(page, 1);
f2fs: fix to unlock page before exit If we got an error after lock_page, we should unlock it before exit. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-20 09:26:09 +08:00
break;
f2fs: use meta_inode cache to improve roll-forward speed Previously, all the dnode pages should be read during the roll-forward recovery. Even worsely, whole the chain was traversed twice. This patch removes that redundant and costly read operations by using page cache of meta_inode and readahead function as well. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 04:49:55 +08:00
}
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
entry = get_fsync_inode(head, ino_of_node(page));
if (!entry)
goto next;
f2fs: fix roll-forward missing scenarios We can summarize the roll forward recovery scenarios as follows. [Term] F: fsync_mark, D: dentry_mark 1. inode(x) | CP | inode(x) | dnode(F) -> Update the latest inode(x). 2. inode(x) | CP | inode(F) | dnode(F) -> No problem. 3. inode(x) | CP | dnode(F) | inode(x) -> Recover to the latest dnode(F), and drop the last inode(x) 4. inode(x) | CP | dnode(F) | inode(F) -> No problem. 5. CP | inode(x) | dnode(F) -> The inode(DF) was missing. Should drop this dnode(F). 6. CP | inode(DF) | dnode(F) -> No problem. 7. CP | dnode(F) | inode(DF) -> If f2fs_iget fails, then goto next to find inode(DF). 8. CP | dnode(F) | inode(x) -> If f2fs_iget fails, then goto next to find inode(DF). But it will fail due to no inode(DF). So, this patch adds some missing points such as #1, #5, #7, and #8. Signed-off-by: Huang Ying <ying.huang@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-16 07:46:08 +08:00
/*
* inode(x) | CP | inode(x) | dnode(F)
* In this case, we can lose the latest inode(x).
f2fs: remove redundant operation during roll-forward recovery If same data is updated multiple times, we don't need to redo whole the operations. Let's just update the lastest one. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 05:29:06 +08:00
* So, call recover_inode for the inode update.
f2fs: fix roll-forward missing scenarios We can summarize the roll forward recovery scenarios as follows. [Term] F: fsync_mark, D: dentry_mark 1. inode(x) | CP | inode(x) | dnode(F) -> Update the latest inode(x). 2. inode(x) | CP | inode(F) | dnode(F) -> No problem. 3. inode(x) | CP | dnode(F) | inode(x) -> Recover to the latest dnode(F), and drop the last inode(x) 4. inode(x) | CP | dnode(F) | inode(F) -> No problem. 5. CP | inode(x) | dnode(F) -> The inode(DF) was missing. Should drop this dnode(F). 6. CP | inode(DF) | dnode(F) -> No problem. 7. CP | dnode(F) | inode(DF) -> If f2fs_iget fails, then goto next to find inode(DF). 8. CP | dnode(F) | inode(x) -> If f2fs_iget fails, then goto next to find inode(DF). But it will fail due to no inode(DF). So, this patch adds some missing points such as #1, #5, #7, and #8. Signed-off-by: Huang Ying <ying.huang@intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-16 07:46:08 +08:00
*/
f2fs: remove redundant operation during roll-forward recovery If same data is updated multiple times, we don't need to redo whole the operations. Let's just update the lastest one. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 05:29:06 +08:00
if (entry->last_inode == blkaddr)
recover_inode(entry->inode, page);
if (entry->last_dentry == blkaddr) {
err = recover_dentry(entry->inode, page);
if (err) {
f2fs_put_page(page, 1);
break;
}
}
f2fs: fix the recovery flow to handle errors correctly We should handle errors during the recovery flow correctly. For example, if we get -ENOMEM, we should report a mount failure instead of conducting the remained mount procedure. Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-20 18:01:06 +08:00
err = do_recover_data(sbi, entry->inode, page, blkaddr);
f2fs: use meta_inode cache to improve roll-forward speed Previously, all the dnode pages should be read during the roll-forward recovery. Even worsely, whole the chain was traversed twice. This patch removes that redundant and costly read operations by using page cache of meta_inode and readahead function as well. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 04:49:55 +08:00
if (err) {
f2fs_put_page(page, 1);
f2fs: fix to unlock page before exit If we got an error after lock_page, we should unlock it before exit. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-05-20 09:26:09 +08:00
break;
f2fs: use meta_inode cache to improve roll-forward speed Previously, all the dnode pages should be read during the roll-forward recovery. Even worsely, whole the chain was traversed twice. This patch removes that redundant and costly read operations by using page cache of meta_inode and readahead function as well. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 04:49:55 +08:00
}
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
if (entry->blkaddr == blkaddr) {
iput(entry->inode);
list_del(&entry->list);
kmem_cache_free(fsync_entry_slab, entry);
}
next:
/* check next segment */
blkaddr = next_blkaddr_of_node(page);
f2fs: use meta_inode cache to improve roll-forward speed Previously, all the dnode pages should be read during the roll-forward recovery. Even worsely, whole the chain was traversed twice. This patch removes that redundant and costly read operations by using page cache of meta_inode and readahead function as well. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 04:49:55 +08:00
f2fs_put_page(page, 1);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
}
f2fs: fix the recovery flow to handle errors correctly We should handle errors during the recovery flow correctly. For example, if we get -ENOMEM, we should report a mount failure instead of conducting the remained mount procedure. Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-20 18:01:06 +08:00
if (!err)
allocate_new_segments(sbi);
return err;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
}
f2fs: fix the recovery flow to handle errors correctly We should handle errors during the recovery flow correctly. For example, if we get -ENOMEM, we should report a mount failure instead of conducting the remained mount procedure. Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-20 18:01:06 +08:00
int recover_fsync_data(struct f2fs_sb_info *sbi)
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
{
f2fs: avoid retrying wrong recovery routine when error was occurred This patch eliminates the propagation of recovery errors to the next mount. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-07-26 06:47:25 +08:00
struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_WARM_NODE);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
struct list_head inode_list;
f2fs: avoid retrying wrong recovery routine when error was occurred This patch eliminates the propagation of recovery errors to the next mount. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-07-26 06:47:25 +08:00
block_t blkaddr;
f2fs: fix the recovery flow to handle errors correctly We should handle errors during the recovery flow correctly. For example, if we get -ENOMEM, we should report a mount failure instead of conducting the remained mount procedure. Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-20 18:01:06 +08:00
int err;
f2fs: use bool for booleans Signed-off-by: Haicheng Li <haicheng.li@linux.intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-23 12:39:32 +08:00
bool need_writecp = false;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
fsync_entry_slab = f2fs_kmem_cache_create("f2fs_fsync_inode_entry",
f2fs: remove the unused ctor argument of f2fs_kmem_cache_create() Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-03-07 18:43:28 +08:00
sizeof(struct fsync_inode_entry));
f2fs: add unlikely() macro for compiler more aggressively This patch adds unlikely() macro into the most of codes. The basic rule is to add that when: - checking unusual errors, - checking page mappings, - and the other unlikely conditions. Change log from v1: - Don't add unlikely for the NULL test and error test: advised by Andi Kleen. Cc: Chao Yu <chao2.yu@samsung.com> Cc: Andi Kleen <andi@firstfloor.org> Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-12-06 14:00:58 +08:00
if (!fsync_entry_slab)
f2fs: fix the recovery flow to handle errors correctly We should handle errors during the recovery flow correctly. For example, if we get -ENOMEM, we should report a mount failure instead of conducting the remained mount procedure. Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-20 18:01:06 +08:00
return -ENOMEM;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
INIT_LIST_HEAD(&inode_list);
/* step #1: find fsynced inode numbers */
f2fs: use bool for booleans Signed-off-by: Haicheng Li <haicheng.li@linux.intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-23 12:39:32 +08:00
sbi->por_doing = true;
f2fs: avoid retrying wrong recovery routine when error was occurred This patch eliminates the propagation of recovery errors to the next mount. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-07-26 06:47:25 +08:00
f2fs: prevent checkpoint during roll-forward Any checkpoint should not be done during the core roll-forward procedure. Especially, it includes error cases too. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-08-14 07:30:46 +08:00
/* prevent checkpoint */
mutex_lock(&sbi->cp_mutex);
f2fs: avoid retrying wrong recovery routine when error was occurred This patch eliminates the propagation of recovery errors to the next mount. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-07-26 06:47:25 +08:00
blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
f2fs: fix the recovery flow to handle errors correctly We should handle errors during the recovery flow correctly. For example, if we get -ENOMEM, we should report a mount failure instead of conducting the remained mount procedure. Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-20 18:01:06 +08:00
err = find_fsync_dnodes(sbi, &inode_list);
if (err)
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
goto out;
if (list_empty(&inode_list))
goto out;
f2fs: use bool for booleans Signed-off-by: Haicheng Li <haicheng.li@linux.intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-23 12:39:32 +08:00
need_writecp = true;
f2fs: remove unneeded write checkpoint in recover_fsync_data Previously, recover_fsync_data still to write checkpoint when there is nothing to recover with normal umount image. It may reduce mount performance and flash memory lifetime, so let's remove it. Signed-off-by: Tan Shu <shu.tan@samsung.com> Signed-off-by: Yu Chao <chao2.yu@samsung.com> Reviewed-by: Gu Zheng <guz.fnst@cn.fujitsu.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-09-24 09:26:24 +08:00
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
/* step #2: recover data */
f2fs: fix the recovery flow to handle errors correctly We should handle errors during the recovery flow correctly. For example, if we get -ENOMEM, we should report a mount failure instead of conducting the remained mount procedure. Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-20 18:01:06 +08:00
err = recover_data(sbi, &inode_list, CURSEG_WARM_NODE);
f2fs: avoid bug_on when error is occurred During the recovery, if an error like EIO or ENOMEM, f2fs_bug_on should skip. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-08-09 01:18:43 +08:00
if (!err)
f2fs: need fsck.f2fs when f2fs_bug_on is triggered If any f2fs_bug_on is triggered, fsck.f2fs is needed. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-03 06:52:58 +08:00
f2fs_bug_on(sbi, !list_empty(&inode_list));
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
out:
f2fs: remove the unused argument "sbi" of func destroy_fsync_dnodes() As destroy_fsync_dnodes() is a simple list-cleanup func, so delete the unused and unrelated f2fs_sb_info argument of it. Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-06-27 09:28:54 +08:00
destroy_fsync_dnodes(&inode_list);
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
kmem_cache_destroy(fsync_entry_slab);
f2fs: avoid retrying wrong recovery routine when error was occurred This patch eliminates the propagation of recovery errors to the next mount. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-07-26 06:47:25 +08:00
f2fs: use meta_inode cache to improve roll-forward speed Previously, all the dnode pages should be read during the roll-forward recovery. Even worsely, whole the chain was traversed twice. This patch removes that redundant and costly read operations by using page cache of meta_inode and readahead function as well. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 04:49:55 +08:00
/* truncate meta pages to be used by the recovery */
truncate_inode_pages_range(META_MAPPING(sbi),
f2fs: check the use of macros on block counts and addresses This patch cleans up the existing and new macros for readability. Rule is like this. ,-----------------------------------------> MAX_BLKADDR -, | ,------------- TOTAL_BLKS ----------------------------, | | | | ,- seg0_blkaddr ,----- sit/nat/ssa/main blkaddress | block | | (SEG0_BLKADDR) | | | | (e.g., MAIN_BLKADDR) | address 0..x................ a b c d ............................. | | global seg# 0...................... m ............................. | | | | `------- MAIN_SEGS -----------' `-------------- TOTAL_SEGS ---------------------------' | | seg# 0..........xx.................. = Note = o GET_SEGNO_FROM_SEG0 : blk address -> global segno o GET_SEGNO : blk address -> segno o START_BLOCK : segno -> starting block address Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-24 02:23:01 +08:00
MAIN_BLKADDR(sbi) << PAGE_CACHE_SHIFT, -1);
f2fs: use meta_inode cache to improve roll-forward speed Previously, all the dnode pages should be read during the roll-forward recovery. Even worsely, whole the chain was traversed twice. This patch removes that redundant and costly read operations by using page cache of meta_inode and readahead function as well. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-12 04:49:55 +08:00
f2fs: avoid retrying wrong recovery routine when error was occurred This patch eliminates the propagation of recovery errors to the next mount. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-07-26 06:47:25 +08:00
if (err) {
truncate_inode_pages_final(NODE_MAPPING(sbi));
truncate_inode_pages_final(META_MAPPING(sbi));
}
f2fs: use bool for booleans Signed-off-by: Haicheng Li <haicheng.li@linux.intel.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-23 12:39:32 +08:00
sbi->por_doing = false;
f2fs: avoid retrying wrong recovery routine when error was occurred This patch eliminates the propagation of recovery errors to the next mount. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-07-26 06:47:25 +08:00
if (err) {
discard_next_dnode(sbi, blkaddr);
/* Flush all the NAT/SIT pages */
while (get_pages(sbi, F2FS_DIRTY_META))
sync_meta_pages(sbi, META, LONG_MAX);
f2fs: prevent checkpoint during roll-forward Any checkpoint should not be done during the core roll-forward procedure. Especially, it includes error cases too. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-08-14 07:30:46 +08:00
set_ckpt_flags(sbi->ckpt, CP_ERROR_FLAG);
mutex_unlock(&sbi->cp_mutex);
f2fs: avoid retrying wrong recovery routine when error was occurred This patch eliminates the propagation of recovery errors to the next mount. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-07-26 06:47:25 +08:00
} else if (need_writecp) {
f2fs: introduce cp_control structure This patch add a new data structure to control checkpoint parameters. Currently, it presents the reason of checkpoint such as is_umount and normal sync. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-21 12:57:51 +08:00
struct cp_control cpc = {
.reason = CP_SYNC,
};
f2fs: prevent checkpoint during roll-forward Any checkpoint should not be done during the core roll-forward procedure. Especially, it includes error cases too. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-08-14 07:30:46 +08:00
mutex_unlock(&sbi->cp_mutex);
f2fs: introduce cp_control structure This patch add a new data structure to control checkpoint parameters. Currently, it presents the reason of checkpoint such as is_umount and normal sync. Reviewed-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-21 12:57:51 +08:00
write_checkpoint(sbi, &cpc);
f2fs: prevent checkpoint during roll-forward Any checkpoint should not be done during the core roll-forward procedure. Especially, it includes error cases too. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-08-14 07:30:46 +08:00
} else {
mutex_unlock(&sbi->cp_mutex);
f2fs: avoid retrying wrong recovery routine when error was occurred This patch eliminates the propagation of recovery errors to the next mount. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-07-26 06:47:25 +08:00
}
f2fs: fix the recovery flow to handle errors correctly We should handle errors during the recovery flow correctly. For example, if we get -ENOMEM, we should report a mount failure instead of conducting the remained mount procedure. Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-03-20 18:01:06 +08:00
return err;
f2fs: add recovery routines for roll-forward This adds roll-forward routines to recover fsynced data. - F2FS uses basically roll-back model with checkpointing. - In order to implement fsync(), there are two approaches as follows. 1. A roll-back model with checkpointing at every fsync() : This is a naive method, but suffers from very low performance. 2. A roll-forward model : F2FS adopts this model where all the fsynced data should be recovered, which were written after checkpointing was done. In order to figure out the data, F2FS keeps a "fsync" mark in direct node blocks. In addition, F2FS remains the location of next node block in each direct node block for reconstructing the chain of node blocks during the recovery. - In order to enhance the performance, F2FS keeps a "dentry" mark also in direct node blocks. If this is set during the recovery, F2FS replays adding a dentry. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 16:13:32 +08:00
}