2012-11-29 04:28:09 +00:00
|
|
|
/*
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
* fs/f2fs/node.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 <linux/mpage.h>
|
|
|
|
#include <linux/backing-dev.h>
|
|
|
|
#include <linux/blkdev.h>
|
|
|
|
#include <linux/pagevec.h>
|
|
|
|
#include <linux/swap.h>
|
|
|
|
|
|
|
|
#include "f2fs.h"
|
|
|
|
#include "node.h"
|
|
|
|
#include "segment.h"
|
2017-07-18 01:48:12 +00:00
|
|
|
#include "xattr.h"
|
2014-12-18 03:58:58 +00:00
|
|
|
#include "trace.h"
|
2013-04-19 16:28:52 +00:00
|
|
|
#include <trace/events/f2fs.h>
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2017-04-08 23:11:36 +00:00
|
|
|
#define on_build_free_nids(nmi) mutex_is_locked(&(nm_i)->build_lock)
|
2014-02-21 10:08:29 +00:00
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
static struct kmem_cache *nat_entry_slab;
|
|
|
|
static struct kmem_cache *free_nid_slab;
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
static struct kmem_cache *nat_entry_set_slab;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2018-04-24 05:02:31 +00:00
|
|
|
/*
|
|
|
|
* Check whether the given nid is within node id range.
|
|
|
|
*/
|
|
|
|
int check_nid_range(struct f2fs_sb_info *sbi, nid_t nid)
|
|
|
|
{
|
|
|
|
if (unlikely(nid < F2FS_ROOT_INO(sbi) || nid >= NM_I(sbi)->max_nid)) {
|
|
|
|
set_sbi_flag(sbi, SBI_NEED_FSCK);
|
|
|
|
f2fs_msg(sbi->sb, KERN_WARNING,
|
|
|
|
"%s: out-of-range nid=%x, run fsck to fix.",
|
|
|
|
__func__, nid);
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2014-04-16 01:47:06 +00:00
|
|
|
bool available_free_memory(struct f2fs_sb_info *sbi, int type)
|
2014-03-19 04:31:37 +00:00
|
|
|
{
|
2014-04-16 01:47:06 +00:00
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
2014-03-19 04:31:37 +00:00
|
|
|
struct sysinfo val;
|
2014-11-06 23:24:46 +00:00
|
|
|
unsigned long avail_ram;
|
2014-03-19 04:31:37 +00:00
|
|
|
unsigned long mem_size = 0;
|
2014-04-16 01:47:06 +00:00
|
|
|
bool res = false;
|
2014-03-19 04:31:37 +00:00
|
|
|
|
|
|
|
si_meminfo(&val);
|
2014-11-06 23:24:46 +00:00
|
|
|
|
|
|
|
/* only uses low memory */
|
|
|
|
avail_ram = val.totalram - val.totalhigh;
|
|
|
|
|
2015-02-05 09:54:31 +00:00
|
|
|
/*
|
|
|
|
* give 25%, 25%, 50%, 50%, 50% memory for each components respectively
|
|
|
|
*/
|
2014-04-16 01:47:06 +00:00
|
|
|
if (type == FREE_NIDS) {
|
2017-09-29 05:59:35 +00:00
|
|
|
mem_size = (nm_i->nid_cnt[FREE_NID] *
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
sizeof(struct free_nid)) >> PAGE_SHIFT;
|
2014-11-06 23:24:46 +00:00
|
|
|
res = mem_size < ((avail_ram * nm_i->ram_thresh / 100) >> 2);
|
2014-04-16 01:47:06 +00:00
|
|
|
} else if (type == NAT_ENTRIES) {
|
2014-11-06 23:24:46 +00:00
|
|
|
mem_size = (nm_i->nat_cnt * sizeof(struct nat_entry)) >>
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
|
|
|
PAGE_SHIFT;
|
2014-11-06 23:24:46 +00:00
|
|
|
res = mem_size < ((avail_ram * nm_i->ram_thresh / 100) >> 2);
|
2016-06-02 22:24:24 +00:00
|
|
|
if (excess_cached_nats(sbi))
|
|
|
|
res = false;
|
2015-10-08 17:40:07 +00:00
|
|
|
} else if (type == DIRTY_DENTS) {
|
|
|
|
if (sbi->sb->s_bdi->wb.dirty_exceeded)
|
|
|
|
return false;
|
|
|
|
mem_size = get_pages(sbi, F2FS_DIRTY_DENTS);
|
|
|
|
res = mem_size < ((avail_ram * nm_i->ram_thresh / 100) >> 1);
|
2014-11-06 23:24:46 +00:00
|
|
|
} else if (type == INO_ENTRIES) {
|
|
|
|
int i;
|
|
|
|
|
2017-09-29 05:59:38 +00:00
|
|
|
for (i = 0; i < MAX_INO_ENTRY; i++)
|
2017-03-18 01:26:13 +00:00
|
|
|
mem_size += sbi->im[i].ino_num *
|
|
|
|
sizeof(struct ino_entry);
|
|
|
|
mem_size >>= PAGE_SHIFT;
|
2014-11-06 23:24:46 +00:00
|
|
|
res = mem_size < ((avail_ram * nm_i->ram_thresh / 100) >> 1);
|
2015-02-05 09:54:31 +00:00
|
|
|
} else if (type == EXTENT_CACHE) {
|
2015-12-22 03:20:15 +00:00
|
|
|
mem_size = (atomic_read(&sbi->total_ext_tree) *
|
|
|
|
sizeof(struct extent_tree) +
|
2015-02-05 09:54:31 +00:00
|
|
|
atomic_read(&sbi->total_ext_node) *
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
|
|
|
sizeof(struct extent_node)) >> PAGE_SHIFT;
|
2015-02-05 09:54:31 +00:00
|
|
|
res = mem_size < ((avail_ram * nm_i->ram_thresh / 100) >> 1);
|
2017-10-19 02:05:57 +00:00
|
|
|
} else if (type == INMEM_PAGES) {
|
|
|
|
/* it allows 20% / total_ram for inmemory pages */
|
|
|
|
mem_size = get_pages(sbi, F2FS_INMEM_PAGES);
|
|
|
|
res = mem_size < (val.totalram / 5);
|
2014-12-09 14:08:59 +00:00
|
|
|
} else {
|
2016-01-10 00:14:08 +00:00
|
|
|
if (!sbi->sb->s_bdi->wb.dirty_exceeded)
|
|
|
|
return true;
|
2014-04-16 01:47:06 +00:00
|
|
|
}
|
|
|
|
return res;
|
2014-03-19 04:31:37 +00:00
|
|
|
}
|
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
static void clear_node_page_dirty(struct page *page)
|
|
|
|
{
|
|
|
|
struct address_space *mapping = page->mapping;
|
|
|
|
unsigned int long flags;
|
|
|
|
|
|
|
|
if (PageDirty(page)) {
|
2018-04-10 23:36:56 +00:00
|
|
|
xa_lock_irqsave(&mapping->i_pages, flags);
|
|
|
|
radix_tree_tag_clear(&mapping->i_pages,
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
page_index(page),
|
|
|
|
PAGECACHE_TAG_DIRTY);
|
2018-04-10 23:36:56 +00:00
|
|
|
xa_unlock_irqrestore(&mapping->i_pages, flags);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
clear_page_dirty_for_io(page);
|
2014-09-02 22:31:18 +00:00
|
|
|
dec_page_count(F2FS_M_SB(mapping), F2FS_DIRTY_NODES);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
ClearPageUptodate(page);
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct page *get_current_nat_page(struct f2fs_sb_info *sbi, nid_t nid)
|
|
|
|
{
|
|
|
|
pgoff_t index = current_nat_addr(sbi, nid);
|
|
|
|
return get_meta_page(sbi, index);
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct page *get_next_nat_page(struct f2fs_sb_info *sbi, nid_t nid)
|
|
|
|
{
|
|
|
|
struct page *src_page;
|
|
|
|
struct page *dst_page;
|
|
|
|
pgoff_t src_off;
|
|
|
|
pgoff_t dst_off;
|
|
|
|
void *src_addr;
|
|
|
|
void *dst_addr;
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
|
|
|
|
src_off = current_nat_addr(sbi, nid);
|
|
|
|
dst_off = next_nat_addr(sbi, src_off);
|
|
|
|
|
|
|
|
/* get current nat block page with lock */
|
|
|
|
src_page = get_meta_page(sbi, src_off);
|
|
|
|
dst_page = grab_meta_page(sbi, dst_off);
|
2014-09-02 22:52:58 +00:00
|
|
|
f2fs_bug_on(sbi, PageDirty(src_page));
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
src_addr = page_address(src_page);
|
|
|
|
dst_addr = page_address(dst_page);
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
|
|
|
memcpy(dst_addr, src_addr, PAGE_SIZE);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
set_page_dirty(dst_page);
|
|
|
|
f2fs_put_page(src_page, 1);
|
|
|
|
|
|
|
|
set_to_next_nat(nm_i, nid);
|
|
|
|
|
|
|
|
return dst_page;
|
|
|
|
}
|
|
|
|
|
2017-11-10 21:36:51 +00:00
|
|
|
static struct nat_entry *__alloc_nat_entry(nid_t nid, bool no_fail)
|
|
|
|
{
|
|
|
|
struct nat_entry *new;
|
|
|
|
|
|
|
|
if (no_fail)
|
2018-01-25 10:57:25 +00:00
|
|
|
new = f2fs_kmem_cache_alloc(nat_entry_slab, GFP_F2FS_ZERO);
|
2017-11-10 21:36:51 +00:00
|
|
|
else
|
2018-01-25 10:57:25 +00:00
|
|
|
new = kmem_cache_alloc(nat_entry_slab, GFP_F2FS_ZERO);
|
2017-11-10 21:36:51 +00:00
|
|
|
if (new) {
|
|
|
|
nat_set_nid(new, nid);
|
|
|
|
nat_reset_flag(new);
|
|
|
|
}
|
|
|
|
return new;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void __free_nat_entry(struct nat_entry *e)
|
|
|
|
{
|
|
|
|
kmem_cache_free(nat_entry_slab, e);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* must be locked by nat_tree_lock */
|
|
|
|
static struct nat_entry *__init_nat_entry(struct f2fs_nm_info *nm_i,
|
|
|
|
struct nat_entry *ne, struct f2fs_nat_entry *raw_ne, bool no_fail)
|
|
|
|
{
|
|
|
|
if (no_fail)
|
|
|
|
f2fs_radix_tree_insert(&nm_i->nat_root, nat_get_nid(ne), ne);
|
|
|
|
else if (radix_tree_insert(&nm_i->nat_root, nat_get_nid(ne), ne))
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
if (raw_ne)
|
|
|
|
node_info_from_raw_nat(&ne->ni, raw_ne);
|
|
|
|
list_add_tail(&ne->list, &nm_i->nat_entries);
|
|
|
|
nm_i->nat_cnt++;
|
|
|
|
return ne;
|
|
|
|
}
|
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
static struct nat_entry *__lookup_nat_cache(struct f2fs_nm_info *nm_i, nid_t n)
|
|
|
|
{
|
|
|
|
return radix_tree_lookup(&nm_i->nat_root, n);
|
|
|
|
}
|
|
|
|
|
|
|
|
static unsigned int __gang_lookup_nat_cache(struct f2fs_nm_info *nm_i,
|
|
|
|
nid_t start, unsigned int nr, struct nat_entry **ep)
|
|
|
|
{
|
|
|
|
return radix_tree_gang_lookup(&nm_i->nat_root, (void **)ep, start, nr);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void __del_from_nat_cache(struct f2fs_nm_info *nm_i, struct nat_entry *e)
|
|
|
|
{
|
|
|
|
list_del(&e->list);
|
|
|
|
radix_tree_delete(&nm_i->nat_root, nat_get_nid(e));
|
|
|
|
nm_i->nat_cnt--;
|
2017-11-10 21:36:51 +00:00
|
|
|
__free_nat_entry(e);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2018-03-20 15:08:30 +00:00
|
|
|
static struct nat_entry_set *__grab_nat_entry_set(struct f2fs_nm_info *nm_i,
|
|
|
|
struct nat_entry *ne)
|
2014-09-22 18:40:48 +00:00
|
|
|
{
|
|
|
|
nid_t set = NAT_BLOCK_OFFSET(ne->ni.nid);
|
|
|
|
struct nat_entry_set *head;
|
|
|
|
|
|
|
|
head = radix_tree_lookup(&nm_i->nat_set_root, set);
|
|
|
|
if (!head) {
|
2015-08-20 15:51:56 +00:00
|
|
|
head = f2fs_kmem_cache_alloc(nat_entry_set_slab, GFP_NOFS);
|
2014-09-22 18:40:48 +00:00
|
|
|
|
|
|
|
INIT_LIST_HEAD(&head->entry_list);
|
|
|
|
INIT_LIST_HEAD(&head->set_list);
|
|
|
|
head->set = set;
|
|
|
|
head->entry_cnt = 0;
|
2014-12-05 18:39:49 +00:00
|
|
|
f2fs_radix_tree_insert(&nm_i->nat_set_root, set, head);
|
2014-09-22 18:40:48 +00:00
|
|
|
}
|
2018-03-20 15:08:30 +00:00
|
|
|
return head;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void __set_nat_cache_dirty(struct f2fs_nm_info *nm_i,
|
|
|
|
struct nat_entry *ne)
|
|
|
|
{
|
|
|
|
struct nat_entry_set *head;
|
|
|
|
bool new_ne = nat_get_blkaddr(ne) == NEW_ADDR;
|
|
|
|
|
|
|
|
if (!new_ne)
|
|
|
|
head = __grab_nat_entry_set(nm_i, ne);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* update entry_cnt in below condition:
|
|
|
|
* 1. update NEW_ADDR to valid block address;
|
|
|
|
* 2. update old block address to new one;
|
|
|
|
*/
|
|
|
|
if (!new_ne && (get_nat_flag(ne, IS_PREALLOC) ||
|
|
|
|
!get_nat_flag(ne, IS_DIRTY)))
|
|
|
|
head->entry_cnt++;
|
|
|
|
|
|
|
|
set_nat_flag(ne, IS_PREALLOC, new_ne);
|
2017-06-05 10:29:08 +00:00
|
|
|
|
|
|
|
if (get_nat_flag(ne, IS_DIRTY))
|
|
|
|
goto refresh_list;
|
|
|
|
|
2014-09-22 18:40:48 +00:00
|
|
|
nm_i->dirty_nat_cnt++;
|
|
|
|
set_nat_flag(ne, IS_DIRTY, true);
|
2017-06-05 10:29:08 +00:00
|
|
|
refresh_list:
|
2018-03-20 15:08:30 +00:00
|
|
|
if (new_ne)
|
2017-06-05 10:29:08 +00:00
|
|
|
list_del_init(&ne->list);
|
|
|
|
else
|
|
|
|
list_move_tail(&ne->list, &head->entry_list);
|
2014-09-22 18:40:48 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void __clear_nat_cache_dirty(struct f2fs_nm_info *nm_i,
|
2017-02-28 13:34:47 +00:00
|
|
|
struct nat_entry_set *set, struct nat_entry *ne)
|
2014-09-22 18:40:48 +00:00
|
|
|
{
|
2017-02-28 13:34:47 +00:00
|
|
|
list_move_tail(&ne->list, &nm_i->nat_entries);
|
|
|
|
set_nat_flag(ne, IS_DIRTY, false);
|
|
|
|
set->entry_cnt--;
|
|
|
|
nm_i->dirty_nat_cnt--;
|
2014-09-22 18:40:48 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static unsigned int __gang_lookup_nat_set(struct f2fs_nm_info *nm_i,
|
|
|
|
nid_t start, unsigned int nr, struct nat_entry_set **ep)
|
|
|
|
{
|
|
|
|
return radix_tree_gang_lookup(&nm_i->nat_set_root, (void **)ep,
|
|
|
|
start, nr);
|
|
|
|
}
|
|
|
|
|
2015-04-30 01:31:19 +00:00
|
|
|
int need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
struct nat_entry *e;
|
2015-04-30 01:31:19 +00:00
|
|
|
bool need = false;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2016-08-04 18:38:25 +00:00
|
|
|
down_read(&nm_i->nat_tree_lock);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
e = __lookup_nat_cache(nm_i, nid);
|
2015-04-30 01:31:19 +00:00
|
|
|
if (e) {
|
|
|
|
if (!get_nat_flag(e, IS_CHECKPOINTED) &&
|
|
|
|
!get_nat_flag(e, HAS_FSYNCED_INODE))
|
|
|
|
need = true;
|
|
|
|
}
|
2016-08-04 18:38:25 +00:00
|
|
|
up_read(&nm_i->nat_tree_lock);
|
2015-04-30 01:31:19 +00:00
|
|
|
return need;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2015-04-30 01:31:19 +00:00
|
|
|
bool is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid)
|
2014-03-20 12:52:53 +00:00
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
struct nat_entry *e;
|
2015-04-30 01:31:19 +00:00
|
|
|
bool is_cp = true;
|
2014-03-20 12:52:53 +00:00
|
|
|
|
2016-08-04 18:38:25 +00:00
|
|
|
down_read(&nm_i->nat_tree_lock);
|
2015-04-30 01:31:19 +00:00
|
|
|
e = __lookup_nat_cache(nm_i, nid);
|
|
|
|
if (e && !get_nat_flag(e, IS_CHECKPOINTED))
|
|
|
|
is_cp = false;
|
2016-08-04 18:38:25 +00:00
|
|
|
up_read(&nm_i->nat_tree_lock);
|
2015-04-30 01:31:19 +00:00
|
|
|
return is_cp;
|
2014-03-20 12:52:53 +00:00
|
|
|
}
|
|
|
|
|
f2fs: fix conditions to remain recovery information in f2fs_sync_file
This patch revisited whole the recovery information during the f2fs_sync_file.
In this patch, there are three information to make a decision.
a) IS_CHECKPOINTED, /* is it checkpointed before? */
b) HAS_FSYNCED_INODE, /* is the inode fsynced before? */
c) HAS_LAST_FSYNC, /* has the latest node fsync mark? */
And, the scenarios for our rule are based on:
[Term] F: fsync_mark, D: dentry_mark
1. inode(x) | CP | inode(x) | dnode(F)
2. inode(x) | CP | inode(F) | dnode(F)
3. inode(x) | CP | dnode(F) | inode(x) | inode(F)
4. inode(x) | CP | dnode(F) | inode(F)
5. CP | inode(x) | dnode(F) | inode(DF)
6. CP | inode(DF) | dnode(F)
7. CP | dnode(F) | inode(DF)
8. CP | dnode(F) | inode(x) | inode(DF)
For example, #3, the three conditions should be changed as follows.
inode(x) | CP | dnode(F) | inode(x) | inode(F)
a) x o o o o
b) x x x x o
c) x o o x o
If f2fs_sync_file stops ------^,
it should write inode(F) --------------^
So, the need_inode_block_update should return true, since
c) get_nat_flag(e, HAS_LAST_FSYNC), is false.
For example, #8,
CP | alloc | dnode(F) | inode(x) | inode(DF)
a) o x x x x
b) x x x o
c) o o x o
If f2fs_sync_file stops -------^,
it should write inode(DF) --------------^
Note that, the roll-forward policy should follow this rule, which means,
if there are any missing blocks, we doesn't need to recover that inode.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-15 21:50:48 +00:00
|
|
|
bool need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino)
|
2014-06-03 15:39:42 +00:00
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
struct nat_entry *e;
|
f2fs: fix conditions to remain recovery information in f2fs_sync_file
This patch revisited whole the recovery information during the f2fs_sync_file.
In this patch, there are three information to make a decision.
a) IS_CHECKPOINTED, /* is it checkpointed before? */
b) HAS_FSYNCED_INODE, /* is the inode fsynced before? */
c) HAS_LAST_FSYNC, /* has the latest node fsync mark? */
And, the scenarios for our rule are based on:
[Term] F: fsync_mark, D: dentry_mark
1. inode(x) | CP | inode(x) | dnode(F)
2. inode(x) | CP | inode(F) | dnode(F)
3. inode(x) | CP | dnode(F) | inode(x) | inode(F)
4. inode(x) | CP | dnode(F) | inode(F)
5. CP | inode(x) | dnode(F) | inode(DF)
6. CP | inode(DF) | dnode(F)
7. CP | dnode(F) | inode(DF)
8. CP | dnode(F) | inode(x) | inode(DF)
For example, #3, the three conditions should be changed as follows.
inode(x) | CP | dnode(F) | inode(x) | inode(F)
a) x o o o o
b) x x x x o
c) x o o x o
If f2fs_sync_file stops ------^,
it should write inode(F) --------------^
So, the need_inode_block_update should return true, since
c) get_nat_flag(e, HAS_LAST_FSYNC), is false.
For example, #8,
CP | alloc | dnode(F) | inode(x) | inode(DF)
a) o x x x x
b) x x x o
c) o o x o
If f2fs_sync_file stops -------^,
it should write inode(DF) --------------^
Note that, the roll-forward policy should follow this rule, which means,
if there are any missing blocks, we doesn't need to recover that inode.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-15 21:50:48 +00:00
|
|
|
bool need_update = true;
|
2014-06-03 15:39:42 +00:00
|
|
|
|
2016-08-04 18:38:25 +00:00
|
|
|
down_read(&nm_i->nat_tree_lock);
|
f2fs: fix conditions to remain recovery information in f2fs_sync_file
This patch revisited whole the recovery information during the f2fs_sync_file.
In this patch, there are three information to make a decision.
a) IS_CHECKPOINTED, /* is it checkpointed before? */
b) HAS_FSYNCED_INODE, /* is the inode fsynced before? */
c) HAS_LAST_FSYNC, /* has the latest node fsync mark? */
And, the scenarios for our rule are based on:
[Term] F: fsync_mark, D: dentry_mark
1. inode(x) | CP | inode(x) | dnode(F)
2. inode(x) | CP | inode(F) | dnode(F)
3. inode(x) | CP | dnode(F) | inode(x) | inode(F)
4. inode(x) | CP | dnode(F) | inode(F)
5. CP | inode(x) | dnode(F) | inode(DF)
6. CP | inode(DF) | dnode(F)
7. CP | dnode(F) | inode(DF)
8. CP | dnode(F) | inode(x) | inode(DF)
For example, #3, the three conditions should be changed as follows.
inode(x) | CP | dnode(F) | inode(x) | inode(F)
a) x o o o o
b) x x x x o
c) x o o x o
If f2fs_sync_file stops ------^,
it should write inode(F) --------------^
So, the need_inode_block_update should return true, since
c) get_nat_flag(e, HAS_LAST_FSYNC), is false.
For example, #8,
CP | alloc | dnode(F) | inode(x) | inode(DF)
a) o x x x x
b) x x x o
c) o o x o
If f2fs_sync_file stops -------^,
it should write inode(DF) --------------^
Note that, the roll-forward policy should follow this rule, which means,
if there are any missing blocks, we doesn't need to recover that inode.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-15 21:50:48 +00:00
|
|
|
e = __lookup_nat_cache(nm_i, ino);
|
|
|
|
if (e && get_nat_flag(e, HAS_LAST_FSYNC) &&
|
|
|
|
(get_nat_flag(e, IS_CHECKPOINTED) ||
|
|
|
|
get_nat_flag(e, HAS_FSYNCED_INODE)))
|
|
|
|
need_update = false;
|
2016-08-04 18:38:25 +00:00
|
|
|
up_read(&nm_i->nat_tree_lock);
|
f2fs: fix conditions to remain recovery information in f2fs_sync_file
This patch revisited whole the recovery information during the f2fs_sync_file.
In this patch, there are three information to make a decision.
a) IS_CHECKPOINTED, /* is it checkpointed before? */
b) HAS_FSYNCED_INODE, /* is the inode fsynced before? */
c) HAS_LAST_FSYNC, /* has the latest node fsync mark? */
And, the scenarios for our rule are based on:
[Term] F: fsync_mark, D: dentry_mark
1. inode(x) | CP | inode(x) | dnode(F)
2. inode(x) | CP | inode(F) | dnode(F)
3. inode(x) | CP | dnode(F) | inode(x) | inode(F)
4. inode(x) | CP | dnode(F) | inode(F)
5. CP | inode(x) | dnode(F) | inode(DF)
6. CP | inode(DF) | dnode(F)
7. CP | dnode(F) | inode(DF)
8. CP | dnode(F) | inode(x) | inode(DF)
For example, #3, the three conditions should be changed as follows.
inode(x) | CP | dnode(F) | inode(x) | inode(F)
a) x o o o o
b) x x x x o
c) x o o x o
If f2fs_sync_file stops ------^,
it should write inode(F) --------------^
So, the need_inode_block_update should return true, since
c) get_nat_flag(e, HAS_LAST_FSYNC), is false.
For example, #8,
CP | alloc | dnode(F) | inode(x) | inode(DF)
a) o x x x x
b) x x x o
c) o o x o
If f2fs_sync_file stops -------^,
it should write inode(DF) --------------^
Note that, the roll-forward policy should follow this rule, which means,
if there are any missing blocks, we doesn't need to recover that inode.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-15 21:50:48 +00:00
|
|
|
return need_update;
|
2014-06-03 15:39:42 +00:00
|
|
|
}
|
|
|
|
|
2017-11-10 21:36:51 +00:00
|
|
|
/* must be locked by nat_tree_lock */
|
2016-02-19 10:12:28 +00:00
|
|
|
static void cache_nat_entry(struct f2fs_sb_info *sbi, nid_t nid,
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
struct f2fs_nat_entry *ne)
|
|
|
|
{
|
2016-02-19 10:12:28 +00:00
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
2017-11-10 21:36:51 +00:00
|
|
|
struct nat_entry *new, *e;
|
2014-12-05 18:39:49 +00:00
|
|
|
|
2017-11-10 21:36:51 +00:00
|
|
|
new = __alloc_nat_entry(nid, false);
|
|
|
|
if (!new)
|
|
|
|
return;
|
|
|
|
|
|
|
|
down_write(&nm_i->nat_tree_lock);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
e = __lookup_nat_cache(nm_i, nid);
|
2017-11-10 21:36:51 +00:00
|
|
|
if (!e)
|
|
|
|
e = __init_nat_entry(nm_i, new, ne, false);
|
|
|
|
else
|
2016-10-11 17:36:12 +00:00
|
|
|
f2fs_bug_on(sbi, nat_get_ino(e) != le32_to_cpu(ne->ino) ||
|
|
|
|
nat_get_blkaddr(e) !=
|
|
|
|
le32_to_cpu(ne->block_addr) ||
|
2016-02-19 10:12:28 +00:00
|
|
|
nat_get_version(e) != ne->version);
|
2017-11-10 21:36:51 +00:00
|
|
|
up_write(&nm_i->nat_tree_lock);
|
|
|
|
if (e != new)
|
|
|
|
__free_nat_entry(new);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void set_node_addr(struct f2fs_sb_info *sbi, struct node_info *ni,
|
2014-03-20 12:52:53 +00:00
|
|
|
block_t new_blkaddr, bool fsync_done)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
struct nat_entry *e;
|
2017-11-10 21:36:51 +00:00
|
|
|
struct nat_entry *new = __alloc_nat_entry(ni->nid, true);
|
2014-12-05 18:39:49 +00:00
|
|
|
|
2016-08-04 18:38:25 +00:00
|
|
|
down_write(&nm_i->nat_tree_lock);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
e = __lookup_nat_cache(nm_i, ni->nid);
|
|
|
|
if (!e) {
|
2017-11-10 21:36:51 +00:00
|
|
|
e = __init_nat_entry(nm_i, new, NULL, true);
|
2014-12-18 09:37:21 +00:00
|
|
|
copy_node_info(&e->ni, ni);
|
2014-09-02 22:52:58 +00:00
|
|
|
f2fs_bug_on(sbi, ni->blk_addr == NEW_ADDR);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
} else if (new_blkaddr == NEW_ADDR) {
|
|
|
|
/*
|
|
|
|
* when nid is reallocated,
|
|
|
|
* previous nat entry can be remained in nat cache.
|
|
|
|
* So, reinitialize it with new information.
|
|
|
|
*/
|
2014-12-18 09:37:21 +00:00
|
|
|
copy_node_info(&e->ni, ni);
|
2014-09-02 22:52:58 +00:00
|
|
|
f2fs_bug_on(sbi, ni->blk_addr != NULL_ADDR);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
2017-11-10 21:36:51 +00:00
|
|
|
/* let's free early to reduce memory consumption */
|
|
|
|
if (e != new)
|
|
|
|
__free_nat_entry(new);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
/* sanity check */
|
2014-09-02 22:52:58 +00:00
|
|
|
f2fs_bug_on(sbi, nat_get_blkaddr(e) != ni->blk_addr);
|
|
|
|
f2fs_bug_on(sbi, nat_get_blkaddr(e) == NULL_ADDR &&
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
new_blkaddr == NULL_ADDR);
|
2014-09-02 22:52:58 +00:00
|
|
|
f2fs_bug_on(sbi, nat_get_blkaddr(e) == NEW_ADDR &&
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
new_blkaddr == NEW_ADDR);
|
2014-09-02 22:52:58 +00:00
|
|
|
f2fs_bug_on(sbi, nat_get_blkaddr(e) != NEW_ADDR &&
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
nat_get_blkaddr(e) != NULL_ADDR &&
|
|
|
|
new_blkaddr == NEW_ADDR);
|
|
|
|
|
2014-08-06 14:22:50 +00:00
|
|
|
/* increment version no as node is removed */
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (nat_get_blkaddr(e) != NEW_ADDR && new_blkaddr == NULL_ADDR) {
|
|
|
|
unsigned char version = nat_get_version(e);
|
|
|
|
nat_set_version(e, inc_node_version(version));
|
|
|
|
}
|
|
|
|
|
|
|
|
/* change address */
|
|
|
|
nat_set_blkaddr(e, new_blkaddr);
|
f2fs: fix conditions to remain recovery information in f2fs_sync_file
This patch revisited whole the recovery information during the f2fs_sync_file.
In this patch, there are three information to make a decision.
a) IS_CHECKPOINTED, /* is it checkpointed before? */
b) HAS_FSYNCED_INODE, /* is the inode fsynced before? */
c) HAS_LAST_FSYNC, /* has the latest node fsync mark? */
And, the scenarios for our rule are based on:
[Term] F: fsync_mark, D: dentry_mark
1. inode(x) | CP | inode(x) | dnode(F)
2. inode(x) | CP | inode(F) | dnode(F)
3. inode(x) | CP | dnode(F) | inode(x) | inode(F)
4. inode(x) | CP | dnode(F) | inode(F)
5. CP | inode(x) | dnode(F) | inode(DF)
6. CP | inode(DF) | dnode(F)
7. CP | dnode(F) | inode(DF)
8. CP | dnode(F) | inode(x) | inode(DF)
For example, #3, the three conditions should be changed as follows.
inode(x) | CP | dnode(F) | inode(x) | inode(F)
a) x o o o o
b) x x x x o
c) x o o x o
If f2fs_sync_file stops ------^,
it should write inode(F) --------------^
So, the need_inode_block_update should return true, since
c) get_nat_flag(e, HAS_LAST_FSYNC), is false.
For example, #8,
CP | alloc | dnode(F) | inode(x) | inode(DF)
a) o x x x x
b) x x x o
c) o o x o
If f2fs_sync_file stops -------^,
it should write inode(DF) --------------^
Note that, the roll-forward policy should follow this rule, which means,
if there are any missing blocks, we doesn't need to recover that inode.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-15 21:50:48 +00:00
|
|
|
if (new_blkaddr == NEW_ADDR || new_blkaddr == NULL_ADDR)
|
|
|
|
set_nat_flag(e, IS_CHECKPOINTED, false);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
__set_nat_cache_dirty(nm_i, e);
|
2014-03-20 12:52:53 +00:00
|
|
|
|
|
|
|
/* update fsync_mark if its inode nat entry is still alive */
|
2015-04-30 10:35:50 +00:00
|
|
|
if (ni->nid != ni->ino)
|
|
|
|
e = __lookup_nat_cache(nm_i, ni->ino);
|
f2fs: fix conditions to remain recovery information in f2fs_sync_file
This patch revisited whole the recovery information during the f2fs_sync_file.
In this patch, there are three information to make a decision.
a) IS_CHECKPOINTED, /* is it checkpointed before? */
b) HAS_FSYNCED_INODE, /* is the inode fsynced before? */
c) HAS_LAST_FSYNC, /* has the latest node fsync mark? */
And, the scenarios for our rule are based on:
[Term] F: fsync_mark, D: dentry_mark
1. inode(x) | CP | inode(x) | dnode(F)
2. inode(x) | CP | inode(F) | dnode(F)
3. inode(x) | CP | dnode(F) | inode(x) | inode(F)
4. inode(x) | CP | dnode(F) | inode(F)
5. CP | inode(x) | dnode(F) | inode(DF)
6. CP | inode(DF) | dnode(F)
7. CP | dnode(F) | inode(DF)
8. CP | dnode(F) | inode(x) | inode(DF)
For example, #3, the three conditions should be changed as follows.
inode(x) | CP | dnode(F) | inode(x) | inode(F)
a) x o o o o
b) x x x x o
c) x o o x o
If f2fs_sync_file stops ------^,
it should write inode(F) --------------^
So, the need_inode_block_update should return true, since
c) get_nat_flag(e, HAS_LAST_FSYNC), is false.
For example, #8,
CP | alloc | dnode(F) | inode(x) | inode(DF)
a) o x x x x
b) x x x o
c) o o x o
If f2fs_sync_file stops -------^,
it should write inode(DF) --------------^
Note that, the roll-forward policy should follow this rule, which means,
if there are any missing blocks, we doesn't need to recover that inode.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-15 21:50:48 +00:00
|
|
|
if (e) {
|
|
|
|
if (fsync_done && ni->nid == ni->ino)
|
|
|
|
set_nat_flag(e, HAS_FSYNCED_INODE, true);
|
|
|
|
set_nat_flag(e, HAS_LAST_FSYNC, fsync_done);
|
|
|
|
}
|
2016-08-04 18:38:25 +00:00
|
|
|
up_write(&nm_i->nat_tree_lock);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2013-10-24 05:19:18 +00:00
|
|
|
int try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
2015-06-19 22:36:07 +00:00
|
|
|
int nr = nr_shrink;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2016-08-04 18:38:25 +00:00
|
|
|
if (!down_write_trylock(&nm_i->nat_tree_lock))
|
|
|
|
return 0;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
while (nr_shrink && !list_empty(&nm_i->nat_entries)) {
|
|
|
|
struct nat_entry *ne;
|
|
|
|
ne = list_first_entry(&nm_i->nat_entries,
|
|
|
|
struct nat_entry, list);
|
|
|
|
__del_from_nat_cache(nm_i, ne);
|
|
|
|
nr_shrink--;
|
|
|
|
}
|
2016-08-04 18:38:25 +00:00
|
|
|
up_write(&nm_i->nat_tree_lock);
|
2015-06-19 22:36:07 +00:00
|
|
|
return nr - nr_shrink;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2012-11-29 04:28:09 +00:00
|
|
|
/*
|
2014-08-06 14:22:50 +00:00
|
|
|
* This function always returns success
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
*/
|
|
|
|
void get_node_info(struct f2fs_sb_info *sbi, nid_t nid, struct node_info *ni)
|
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA);
|
f2fs: split journal cache from curseg cache
In curseg cache, f2fs caches two different parts:
- datas of current summay block, i.e. summary entries, footer info.
- journal info, i.e. sparse nat/sit entries or io stat info.
With this approach, 1) it may cause higher lock contention when we access
or update both of the parts of cache since we use the same mutex lock
curseg_mutex to protect the cache. 2) current summary block with last
journal info will be writebacked into device as a normal summary block
when flushing, however, we treat journal info as valid one only in current
summary, so most normal summary blocks contain junk journal data, it wastes
remaining space of summary block.
So, in order to fix above issues, we split curseg cache into two parts:
a) current summary block, protected by original mutex lock curseg_mutex
b) journal cache, protected by newly introduced r/w semaphore journal_rwsem
When loading curseg cache during ->mount, we store summary info and
journal info into different caches; When doing checkpoint, we combine
datas of two cache into current summary block for persisting.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-19 10:08:46 +00:00
|
|
|
struct f2fs_journal *journal = curseg->journal;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
nid_t start_nid = START_NID(nid);
|
|
|
|
struct f2fs_nat_block *nat_blk;
|
|
|
|
struct page *page = NULL;
|
|
|
|
struct f2fs_nat_entry ne;
|
|
|
|
struct nat_entry *e;
|
2017-04-22 10:06:26 +00:00
|
|
|
pgoff_t index;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
int i;
|
|
|
|
|
|
|
|
ni->nid = nid;
|
|
|
|
|
|
|
|
/* Check nat cache */
|
2016-08-04 18:38:25 +00:00
|
|
|
down_read(&nm_i->nat_tree_lock);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
e = __lookup_nat_cache(nm_i, nid);
|
|
|
|
if (e) {
|
|
|
|
ni->ino = nat_get_ino(e);
|
|
|
|
ni->blk_addr = nat_get_blkaddr(e);
|
|
|
|
ni->version = nat_get_version(e);
|
2016-08-04 18:38:25 +00:00
|
|
|
up_read(&nm_i->nat_tree_lock);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return;
|
2016-02-19 10:12:28 +00:00
|
|
|
}
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2014-12-31 07:08:26 +00:00
|
|
|
memset(&ne, 0, sizeof(struct f2fs_nat_entry));
|
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
/* Check current segment summary */
|
f2fs: split journal cache from curseg cache
In curseg cache, f2fs caches two different parts:
- datas of current summay block, i.e. summary entries, footer info.
- journal info, i.e. sparse nat/sit entries or io stat info.
With this approach, 1) it may cause higher lock contention when we access
or update both of the parts of cache since we use the same mutex lock
curseg_mutex to protect the cache. 2) current summary block with last
journal info will be writebacked into device as a normal summary block
when flushing, however, we treat journal info as valid one only in current
summary, so most normal summary blocks contain junk journal data, it wastes
remaining space of summary block.
So, in order to fix above issues, we split curseg cache into two parts:
a) current summary block, protected by original mutex lock curseg_mutex
b) journal cache, protected by newly introduced r/w semaphore journal_rwsem
When loading curseg cache during ->mount, we store summary info and
journal info into different caches; When doing checkpoint, we combine
datas of two cache into current summary block for persisting.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-19 10:08:46 +00:00
|
|
|
down_read(&curseg->journal_rwsem);
|
2016-02-14 10:50:40 +00:00
|
|
|
i = lookup_journal_in_cursum(journal, NAT_JOURNAL, nid, 0);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (i >= 0) {
|
2016-02-14 10:50:40 +00:00
|
|
|
ne = nat_in_journal(journal, i);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
node_info_from_raw_nat(ni, &ne);
|
|
|
|
}
|
f2fs: split journal cache from curseg cache
In curseg cache, f2fs caches two different parts:
- datas of current summay block, i.e. summary entries, footer info.
- journal info, i.e. sparse nat/sit entries or io stat info.
With this approach, 1) it may cause higher lock contention when we access
or update both of the parts of cache since we use the same mutex lock
curseg_mutex to protect the cache. 2) current summary block with last
journal info will be writebacked into device as a normal summary block
when flushing, however, we treat journal info as valid one only in current
summary, so most normal summary blocks contain junk journal data, it wastes
remaining space of summary block.
So, in order to fix above issues, we split curseg cache into two parts:
a) current summary block, protected by original mutex lock curseg_mutex
b) journal cache, protected by newly introduced r/w semaphore journal_rwsem
When loading curseg cache during ->mount, we store summary info and
journal info into different caches; When doing checkpoint, we combine
datas of two cache into current summary block for persisting.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-19 10:08:46 +00:00
|
|
|
up_read(&curseg->journal_rwsem);
|
2017-04-22 10:06:26 +00:00
|
|
|
if (i >= 0) {
|
|
|
|
up_read(&nm_i->nat_tree_lock);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
goto cache;
|
2017-04-22 10:06:26 +00:00
|
|
|
}
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
/* Fill node_info from nat page */
|
2017-04-22 10:06:26 +00:00
|
|
|
index = current_nat_addr(sbi, nid);
|
|
|
|
up_read(&nm_i->nat_tree_lock);
|
|
|
|
|
|
|
|
page = get_meta_page(sbi, index);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
nat_blk = (struct f2fs_nat_block *)page_address(page);
|
|
|
|
ne = nat_blk->entries[nid - start_nid];
|
|
|
|
node_info_from_raw_nat(ni, &ne);
|
|
|
|
f2fs_put_page(page, 1);
|
|
|
|
cache:
|
|
|
|
/* cache nat entry */
|
2016-02-19 10:12:28 +00:00
|
|
|
cache_nat_entry(sbi, nid, &ne);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2016-05-06 23:19:43 +00:00
|
|
|
/*
|
|
|
|
* readahead MAX_RA_NODE number of node pages.
|
|
|
|
*/
|
|
|
|
static void ra_node_pages(struct page *parent, int start, int n)
|
|
|
|
{
|
|
|
|
struct f2fs_sb_info *sbi = F2FS_P_SB(parent);
|
|
|
|
struct blk_plug plug;
|
|
|
|
int i, end;
|
|
|
|
nid_t nid;
|
|
|
|
|
|
|
|
blk_start_plug(&plug);
|
|
|
|
|
|
|
|
/* Then, try readahead for siblings of the desired node */
|
|
|
|
end = start + n;
|
|
|
|
end = min(end, NIDS_PER_BLOCK);
|
|
|
|
for (i = start; i < end; i++) {
|
|
|
|
nid = get_nid(parent, i, false);
|
|
|
|
ra_node_page(sbi, nid);
|
|
|
|
}
|
|
|
|
|
|
|
|
blk_finish_plug(&plug);
|
|
|
|
}
|
|
|
|
|
2016-01-26 07:40:44 +00:00
|
|
|
pgoff_t get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs)
|
|
|
|
{
|
|
|
|
const long direct_index = ADDRS_PER_INODE(dn->inode);
|
|
|
|
const long direct_blks = ADDRS_PER_BLOCK;
|
|
|
|
const long indirect_blks = ADDRS_PER_BLOCK * NIDS_PER_BLOCK;
|
|
|
|
unsigned int skipped_unit = ADDRS_PER_BLOCK;
|
|
|
|
int cur_level = dn->cur_level;
|
|
|
|
int max_level = dn->max_level;
|
|
|
|
pgoff_t base = 0;
|
|
|
|
|
|
|
|
if (!dn->max_level)
|
|
|
|
return pgofs + 1;
|
|
|
|
|
|
|
|
while (max_level-- > cur_level)
|
|
|
|
skipped_unit *= NIDS_PER_BLOCK;
|
|
|
|
|
|
|
|
switch (dn->max_level) {
|
|
|
|
case 3:
|
|
|
|
base += 2 * indirect_blks;
|
|
|
|
case 2:
|
|
|
|
base += 2 * direct_blks;
|
|
|
|
case 1:
|
|
|
|
base += direct_index;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
f2fs_bug_on(F2FS_I_SB(dn->inode), 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
return ((pgofs - base) / skipped_unit + 1) * skipped_unit + base;
|
|
|
|
}
|
|
|
|
|
2012-11-29 04:28:09 +00:00
|
|
|
/*
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
* The maximum depth is four.
|
|
|
|
* Offset[0] will have raw inode offset.
|
|
|
|
*/
|
2016-01-26 07:39:35 +00:00
|
|
|
static int get_node_path(struct inode *inode, long block,
|
2013-08-12 12:08:03 +00:00
|
|
|
int offset[4], unsigned int noffset[4])
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
2016-01-26 07:39:35 +00:00
|
|
|
const long direct_index = ADDRS_PER_INODE(inode);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
const long direct_blks = ADDRS_PER_BLOCK;
|
|
|
|
const long dptrs_per_blk = NIDS_PER_BLOCK;
|
|
|
|
const long indirect_blks = ADDRS_PER_BLOCK * NIDS_PER_BLOCK;
|
|
|
|
const long dindirect_blks = indirect_blks * NIDS_PER_BLOCK;
|
|
|
|
int n = 0;
|
|
|
|
int level = 0;
|
|
|
|
|
|
|
|
noffset[0] = 0;
|
|
|
|
|
|
|
|
if (block < direct_index) {
|
2013-03-02 03:41:31 +00:00
|
|
|
offset[n] = block;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
goto got;
|
|
|
|
}
|
|
|
|
block -= direct_index;
|
|
|
|
if (block < direct_blks) {
|
|
|
|
offset[n++] = NODE_DIR1_BLOCK;
|
|
|
|
noffset[n] = 1;
|
2013-03-02 03:41:31 +00:00
|
|
|
offset[n] = block;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
level = 1;
|
|
|
|
goto got;
|
|
|
|
}
|
|
|
|
block -= direct_blks;
|
|
|
|
if (block < direct_blks) {
|
|
|
|
offset[n++] = NODE_DIR2_BLOCK;
|
|
|
|
noffset[n] = 2;
|
2013-03-02 03:41:31 +00:00
|
|
|
offset[n] = block;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
level = 1;
|
|
|
|
goto got;
|
|
|
|
}
|
|
|
|
block -= direct_blks;
|
|
|
|
if (block < indirect_blks) {
|
|
|
|
offset[n++] = NODE_IND1_BLOCK;
|
|
|
|
noffset[n] = 3;
|
|
|
|
offset[n++] = block / direct_blks;
|
|
|
|
noffset[n] = 4 + offset[n - 1];
|
2013-03-02 03:41:31 +00:00
|
|
|
offset[n] = block % direct_blks;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
level = 2;
|
|
|
|
goto got;
|
|
|
|
}
|
|
|
|
block -= indirect_blks;
|
|
|
|
if (block < indirect_blks) {
|
|
|
|
offset[n++] = NODE_IND2_BLOCK;
|
|
|
|
noffset[n] = 4 + dptrs_per_blk;
|
|
|
|
offset[n++] = block / direct_blks;
|
|
|
|
noffset[n] = 5 + dptrs_per_blk + offset[n - 1];
|
2013-03-02 03:41:31 +00:00
|
|
|
offset[n] = block % direct_blks;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
level = 2;
|
|
|
|
goto got;
|
|
|
|
}
|
|
|
|
block -= indirect_blks;
|
|
|
|
if (block < dindirect_blks) {
|
|
|
|
offset[n++] = NODE_DIND_BLOCK;
|
|
|
|
noffset[n] = 5 + (dptrs_per_blk * 2);
|
|
|
|
offset[n++] = block / indirect_blks;
|
|
|
|
noffset[n] = 6 + (dptrs_per_blk * 2) +
|
|
|
|
offset[n - 1] * (dptrs_per_blk + 1);
|
|
|
|
offset[n++] = (block / direct_blks) % dptrs_per_blk;
|
|
|
|
noffset[n] = 7 + (dptrs_per_blk * 2) +
|
|
|
|
offset[n - 2] * (dptrs_per_blk + 1) +
|
|
|
|
offset[n - 1];
|
2013-03-02 03:41:31 +00:00
|
|
|
offset[n] = block % direct_blks;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
level = 3;
|
|
|
|
goto got;
|
|
|
|
} else {
|
2017-08-21 20:51:32 +00:00
|
|
|
return -E2BIG;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
got:
|
|
|
|
return level;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Caller should call f2fs_put_dnode(dn).
|
2013-12-21 10:02:14 +00:00
|
|
|
* Also, it should grab and release a rwsem by calling f2fs_lock_op() and
|
|
|
|
* f2fs_unlock_op() only if ro is not set RDONLY_NODE.
|
f2fs: introduce a new global lock scheme
In the previous version, f2fs uses global locks according to the usage types,
such as directory operations, block allocation, block write, and so on.
Reference the following lock types in f2fs.h.
enum lock_type {
RENAME, /* for renaming operations */
DENTRY_OPS, /* for directory operations */
DATA_WRITE, /* for data write */
DATA_NEW, /* for data allocation */
DATA_TRUNC, /* for data truncate */
NODE_NEW, /* for node allocation */
NODE_TRUNC, /* for node truncate */
NODE_WRITE, /* for node write */
NR_LOCK_TYPE,
};
In that case, we lose the performance under the multi-threading environment,
since every types of operations must be conducted one at a time.
In order to address the problem, let's share the locks globally with a mutex
array regardless of any types.
So, let users grab a mutex and perform their jobs in parallel as much as
possbile.
For this, I propose a new global lock scheme as follows.
0. Data structure
- f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS]
- f2fs_sb_info -> node_write
1. mutex_lock_op(sbi)
- try to get an avaiable lock from the array.
- returns the index of the gottern lock variable.
2. mutex_unlock_op(sbi, index of the lock)
- unlock the given index of the lock.
3. mutex_lock_all(sbi)
- grab all the locks in the array before the checkpoint.
4. mutex_unlock_all(sbi)
- release all the locks in the array after checkpoint.
5. block_operations()
- call mutex_lock_all()
- sync_dirty_dir_inodes()
- grab node_write
- sync_node_pages()
Note that,
the pairs of mutex_lock_op()/mutex_unlock_op() and
mutex_lock_all()/mutex_unlock_all() should be used together.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
|
|
|
* In the case of RDONLY_NODE, we don't need to care about mutex.
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
*/
|
2013-02-26 04:10:46 +00:00
|
|
|
int get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
2014-09-02 22:31:18 +00:00
|
|
|
struct f2fs_sb_info *sbi = F2FS_I_SB(dn->inode);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
struct page *npage[4];
|
2015-02-11 19:25:11 +00:00
|
|
|
struct page *parent = NULL;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
int offset[4];
|
|
|
|
unsigned int noffset[4];
|
|
|
|
nid_t nids[4];
|
2016-01-26 07:40:44 +00:00
|
|
|
int level, i = 0;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
int err = 0;
|
|
|
|
|
2016-01-26 07:39:35 +00:00
|
|
|
level = get_node_path(dn->inode, index, offset, noffset);
|
2017-08-21 20:51:32 +00:00
|
|
|
if (level < 0)
|
|
|
|
return level;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
nids[0] = dn->inode->i_ino;
|
2013-05-20 00:42:28 +00:00
|
|
|
npage[0] = dn->inode_page;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2013-05-20 00:42:28 +00:00
|
|
|
if (!npage[0]) {
|
|
|
|
npage[0] = get_node_page(sbi, nids[0]);
|
|
|
|
if (IS_ERR(npage[0]))
|
|
|
|
return PTR_ERR(npage[0]);
|
|
|
|
}
|
2015-02-11 19:25:11 +00:00
|
|
|
|
|
|
|
/* if inline_data is set, should not report any block indices */
|
|
|
|
if (f2fs_has_inline_data(dn->inode) && index) {
|
2015-03-03 00:28:16 +00:00
|
|
|
err = -ENOENT;
|
2015-02-11 19:25:11 +00:00
|
|
|
f2fs_put_page(npage[0], 1);
|
|
|
|
goto release_out;
|
|
|
|
}
|
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
parent = npage[0];
|
2013-02-19 22:47:06 +00:00
|
|
|
if (level != 0)
|
|
|
|
nids[1] = get_nid(parent, offset[0], true);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
dn->inode_page = npage[0];
|
|
|
|
dn->inode_page_locked = true;
|
|
|
|
|
|
|
|
/* get indirect or direct nodes */
|
|
|
|
for (i = 1; i <= level; i++) {
|
|
|
|
bool done = false;
|
|
|
|
|
2013-02-26 04:10:46 +00:00
|
|
|
if (!nids[i] && mode == ALLOC_NODE) {
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
/* alloc new node */
|
|
|
|
if (!alloc_nid(sbi, &(nids[i]))) {
|
|
|
|
err = -ENOSPC;
|
|
|
|
goto release_pages;
|
|
|
|
}
|
|
|
|
|
|
|
|
dn->nid = nids[i];
|
2017-07-17 11:16:11 +00:00
|
|
|
npage[i] = new_node_page(dn, noffset[i]);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (IS_ERR(npage[i])) {
|
|
|
|
alloc_nid_failed(sbi, nids[i]);
|
|
|
|
err = PTR_ERR(npage[i]);
|
|
|
|
goto release_pages;
|
|
|
|
}
|
|
|
|
|
|
|
|
set_nid(parent, offset[i - 1], nids[i], i == 1);
|
|
|
|
alloc_nid_done(sbi, nids[i]);
|
|
|
|
done = true;
|
2013-02-26 04:10:46 +00:00
|
|
|
} else if (mode == LOOKUP_NODE_RA && i == level && level > 1) {
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
npage[i] = get_node_page_ra(parent, offset[i - 1]);
|
|
|
|
if (IS_ERR(npage[i])) {
|
|
|
|
err = PTR_ERR(npage[i]);
|
|
|
|
goto release_pages;
|
|
|
|
}
|
|
|
|
done = true;
|
|
|
|
}
|
|
|
|
if (i == 1) {
|
|
|
|
dn->inode_page_locked = false;
|
|
|
|
unlock_page(parent);
|
|
|
|
} else {
|
|
|
|
f2fs_put_page(parent, 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!done) {
|
|
|
|
npage[i] = get_node_page(sbi, nids[i]);
|
|
|
|
if (IS_ERR(npage[i])) {
|
|
|
|
err = PTR_ERR(npage[i]);
|
|
|
|
f2fs_put_page(npage[0], 0);
|
|
|
|
goto release_out;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (i < level) {
|
|
|
|
parent = npage[i];
|
|
|
|
nids[i + 1] = get_nid(parent, offset[i], false);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
dn->nid = nids[level];
|
|
|
|
dn->ofs_in_node = offset[level];
|
|
|
|
dn->node_page = npage[level];
|
2017-07-18 16:19:06 +00:00
|
|
|
dn->data_blkaddr = datablock_addr(dn->inode,
|
|
|
|
dn->node_page, dn->ofs_in_node);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
release_pages:
|
|
|
|
f2fs_put_page(parent, 1);
|
|
|
|
if (i > 1)
|
|
|
|
f2fs_put_page(npage[0], 0);
|
|
|
|
release_out:
|
|
|
|
dn->inode_page = NULL;
|
|
|
|
dn->node_page = NULL;
|
2016-01-26 07:40:44 +00:00
|
|
|
if (err == -ENOENT) {
|
|
|
|
dn->cur_level = i;
|
|
|
|
dn->max_level = level;
|
2016-07-09 00:42:21 +00:00
|
|
|
dn->ofs_in_node = offset[level];
|
2016-01-26 07:40:44 +00:00
|
|
|
}
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void truncate_node(struct dnode_of_data *dn)
|
|
|
|
{
|
2014-09-02 22:31:18 +00:00
|
|
|
struct f2fs_sb_info *sbi = F2FS_I_SB(dn->inode);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
struct node_info ni;
|
|
|
|
|
|
|
|
get_node_info(sbi, dn->nid, &ni);
|
|
|
|
|
|
|
|
/* Deallocate node address */
|
2012-12-20 06:10:06 +00:00
|
|
|
invalidate_blocks(sbi, ni.blk_addr);
|
f2fs: don't count inode block in in-memory inode.i_blocks
Previously, we count all inode consumed blocks including inode block,
xattr block, index block, data block into i_blocks, for other generic
filesystems, they won't count inode block into i_blocks, so for
userspace applications or quota system, they may detect incorrect block
count according to i_blocks value in inode.
This patch changes to count all blocks into inode.i_blocks excluding
inode block, for on-disk i_blocks, we keep counting inode block for
backward compatibility.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-05 17:11:31 +00:00
|
|
|
dec_valid_node_count(sbi, dn->inode, dn->nid == dn->inode->i_ino);
|
2014-03-20 12:52:53 +00:00
|
|
|
set_node_addr(sbi, &ni, NULL_ADDR, false);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
if (dn->nid == dn->inode->i_ino) {
|
|
|
|
remove_orphan_inode(sbi, dn->nid);
|
|
|
|
dec_valid_inode_count(sbi);
|
2016-05-20 18:10:10 +00:00
|
|
|
f2fs_inode_synced(dn->inode);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
f2fs: don't count inode block in in-memory inode.i_blocks
Previously, we count all inode consumed blocks including inode block,
xattr block, index block, data block into i_blocks, for other generic
filesystems, they won't count inode block into i_blocks, so for
userspace applications or quota system, they may detect incorrect block
count according to i_blocks value in inode.
This patch changes to count all blocks into inode.i_blocks excluding
inode block, for on-disk i_blocks, we keep counting inode block for
backward compatibility.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-05 17:11:31 +00:00
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
clear_node_page_dirty(dn->node_page);
|
2015-01-28 09:48:42 +00:00
|
|
|
set_sbi_flag(sbi, SBI_IS_DIRTY);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
f2fs_put_page(dn->node_page, 1);
|
2014-01-22 11:41:57 +00:00
|
|
|
|
|
|
|
invalidate_mapping_pages(NODE_MAPPING(sbi),
|
|
|
|
dn->node_page->index, dn->node_page->index);
|
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
dn->node_page = NULL;
|
2013-04-19 16:28:52 +00:00
|
|
|
trace_f2fs_truncate_node(dn->inode, dn->nid, ni.blk_addr);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int truncate_dnode(struct dnode_of_data *dn)
|
|
|
|
{
|
|
|
|
struct page *page;
|
|
|
|
|
|
|
|
if (dn->nid == 0)
|
|
|
|
return 1;
|
|
|
|
|
|
|
|
/* get direct node */
|
2014-09-02 22:31:18 +00:00
|
|
|
page = get_node_page(F2FS_I_SB(dn->inode), dn->nid);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (IS_ERR(page) && PTR_ERR(page) == -ENOENT)
|
|
|
|
return 1;
|
|
|
|
else if (IS_ERR(page))
|
|
|
|
return PTR_ERR(page);
|
|
|
|
|
|
|
|
/* Make dnode_of_data for parameter */
|
|
|
|
dn->node_page = page;
|
|
|
|
dn->ofs_in_node = 0;
|
|
|
|
truncate_data_blocks(dn);
|
|
|
|
truncate_node(dn);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int truncate_nodes(struct dnode_of_data *dn, unsigned int nofs,
|
|
|
|
int ofs, int depth)
|
|
|
|
{
|
|
|
|
struct dnode_of_data rdn = *dn;
|
|
|
|
struct page *page;
|
|
|
|
struct f2fs_node *rn;
|
|
|
|
nid_t child_nid;
|
|
|
|
unsigned int child_nofs;
|
|
|
|
int freed = 0;
|
|
|
|
int i, ret;
|
|
|
|
|
|
|
|
if (dn->nid == 0)
|
|
|
|
return NIDS_PER_BLOCK + 1;
|
|
|
|
|
2013-04-19 16:28:52 +00:00
|
|
|
trace_f2fs_truncate_nodes_enter(dn->inode, dn->nid, dn->data_blkaddr);
|
|
|
|
|
2014-09-02 22:31:18 +00:00
|
|
|
page = get_node_page(F2FS_I_SB(dn->inode), dn->nid);
|
2013-04-19 16:28:52 +00:00
|
|
|
if (IS_ERR(page)) {
|
|
|
|
trace_f2fs_truncate_nodes_exit(dn->inode, PTR_ERR(page));
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return PTR_ERR(page);
|
2013-04-19 16:28:52 +00:00
|
|
|
}
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2016-05-06 23:19:43 +00:00
|
|
|
ra_node_pages(page, ofs, NIDS_PER_BLOCK);
|
|
|
|
|
2013-07-15 09:57:38 +00:00
|
|
|
rn = F2FS_NODE(page);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (depth < 3) {
|
|
|
|
for (i = ofs; i < NIDS_PER_BLOCK; i++, freed++) {
|
|
|
|
child_nid = le32_to_cpu(rn->in.nid[i]);
|
|
|
|
if (child_nid == 0)
|
|
|
|
continue;
|
|
|
|
rdn.nid = child_nid;
|
|
|
|
ret = truncate_dnode(&rdn);
|
|
|
|
if (ret < 0)
|
|
|
|
goto out_err;
|
2016-01-07 21:23:12 +00:00
|
|
|
if (set_nid(page, i, 0, false))
|
|
|
|
dn->node_changed = true;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
} else {
|
|
|
|
child_nofs = nofs + ofs * (NIDS_PER_BLOCK + 1) + 1;
|
|
|
|
for (i = ofs; i < NIDS_PER_BLOCK; i++) {
|
|
|
|
child_nid = le32_to_cpu(rn->in.nid[i]);
|
|
|
|
if (child_nid == 0) {
|
|
|
|
child_nofs += NIDS_PER_BLOCK + 1;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
rdn.nid = child_nid;
|
|
|
|
ret = truncate_nodes(&rdn, child_nofs, 0, depth - 1);
|
|
|
|
if (ret == (NIDS_PER_BLOCK + 1)) {
|
2016-01-07 21:23:12 +00:00
|
|
|
if (set_nid(page, i, 0, false))
|
|
|
|
dn->node_changed = true;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
child_nofs += ret;
|
|
|
|
} else if (ret < 0 && ret != -ENOENT) {
|
|
|
|
goto out_err;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
freed = child_nofs;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!ofs) {
|
|
|
|
/* remove current indirect node */
|
|
|
|
dn->node_page = page;
|
|
|
|
truncate_node(dn);
|
|
|
|
freed++;
|
|
|
|
} else {
|
|
|
|
f2fs_put_page(page, 1);
|
|
|
|
}
|
2013-04-19 16:28:52 +00:00
|
|
|
trace_f2fs_truncate_nodes_exit(dn->inode, freed);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return freed;
|
|
|
|
|
|
|
|
out_err:
|
|
|
|
f2fs_put_page(page, 1);
|
2013-04-19 16:28:52 +00:00
|
|
|
trace_f2fs_truncate_nodes_exit(dn->inode, ret);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int truncate_partial_nodes(struct dnode_of_data *dn,
|
|
|
|
struct f2fs_inode *ri, int *offset, int depth)
|
|
|
|
{
|
|
|
|
struct page *pages[2];
|
|
|
|
nid_t nid[3];
|
|
|
|
nid_t child_nid;
|
|
|
|
int err = 0;
|
|
|
|
int i;
|
|
|
|
int idx = depth - 2;
|
|
|
|
|
|
|
|
nid[0] = le32_to_cpu(ri->i_nid[offset[0] - NODE_DIR1_BLOCK]);
|
|
|
|
if (!nid[0])
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
/* get indirect nodes in the path */
|
f2fs: fix truncate_partial_nodes bug
The truncate_partial_nodes puts pages incorrectly in the following two cases.
Note that the value for argc 'depth' can only be 2 or 3.
Please see truncate_inode_blocks() and truncate_partial_nodes().
1) An err is occurred in the first 'for' loop
When err is occurred with depth = 2, pages[0] is invalid, so this page doesn't
need to be put. There is no problem, however, when depth is 3, it doesn't put
the pages correctly where pages[0] is valid and pages[1] is invalid.
In this case, depth is set to 2 (ref to statemnt depth = i + 1), and then
'goto fail'.
In label 'fail', for (i = depth - 3; i >= 0; i--) cannot meet the condition
because i = -1, so pages[0] cann't be put.
2) An err happened in the second 'for' loop
Now we've got pages[0] with depth = 2, or we've got pages[0] and pages[1]
with depth = 3. When an err is detected, we need 'goto fail' to put such
the pages.
When depth is 2, in label 'fail', for (i = depth - 3; i >= 0; i--) cann't
meet the condition because i = -1, so pages[0] cann't be put.
When depth is 3, in label 'fail', for (i = depth - 3; i >= 0; i--) can
only put pages[0], pages[1] also cann't be put.
Note that 'depth' has been changed before first 'goto fail' (ref to statemnt
depth = i + 1), so passing this modified 'depth' to the tracepoint,
trace_f2fs_truncate_partial_nodes, is also incorrect.
Signed-off-by: Shifei Ge <shifei10.ge@samsung.com>
[Jaegeuk Kim: modify the description and fix one bug]
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-29 07:32:34 +00:00
|
|
|
for (i = 0; i < idx + 1; i++) {
|
2014-08-06 14:22:50 +00:00
|
|
|
/* reference count'll be increased */
|
2014-09-02 22:31:18 +00:00
|
|
|
pages[i] = get_node_page(F2FS_I_SB(dn->inode), nid[i]);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (IS_ERR(pages[i])) {
|
|
|
|
err = PTR_ERR(pages[i]);
|
f2fs: fix truncate_partial_nodes bug
The truncate_partial_nodes puts pages incorrectly in the following two cases.
Note that the value for argc 'depth' can only be 2 or 3.
Please see truncate_inode_blocks() and truncate_partial_nodes().
1) An err is occurred in the first 'for' loop
When err is occurred with depth = 2, pages[0] is invalid, so this page doesn't
need to be put. There is no problem, however, when depth is 3, it doesn't put
the pages correctly where pages[0] is valid and pages[1] is invalid.
In this case, depth is set to 2 (ref to statemnt depth = i + 1), and then
'goto fail'.
In label 'fail', for (i = depth - 3; i >= 0; i--) cannot meet the condition
because i = -1, so pages[0] cann't be put.
2) An err happened in the second 'for' loop
Now we've got pages[0] with depth = 2, or we've got pages[0] and pages[1]
with depth = 3. When an err is detected, we need 'goto fail' to put such
the pages.
When depth is 2, in label 'fail', for (i = depth - 3; i >= 0; i--) cann't
meet the condition because i = -1, so pages[0] cann't be put.
When depth is 3, in label 'fail', for (i = depth - 3; i >= 0; i--) can
only put pages[0], pages[1] also cann't be put.
Note that 'depth' has been changed before first 'goto fail' (ref to statemnt
depth = i + 1), so passing this modified 'depth' to the tracepoint,
trace_f2fs_truncate_partial_nodes, is also incorrect.
Signed-off-by: Shifei Ge <shifei10.ge@samsung.com>
[Jaegeuk Kim: modify the description and fix one bug]
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-29 07:32:34 +00:00
|
|
|
idx = i - 1;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
nid[i + 1] = get_nid(pages[i], offset[i + 1], false);
|
|
|
|
}
|
|
|
|
|
2016-05-06 23:19:43 +00:00
|
|
|
ra_node_pages(pages[idx], offset[idx + 1], NIDS_PER_BLOCK);
|
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
/* free direct nodes linked to a partial indirect node */
|
f2fs: fix truncate_partial_nodes bug
The truncate_partial_nodes puts pages incorrectly in the following two cases.
Note that the value for argc 'depth' can only be 2 or 3.
Please see truncate_inode_blocks() and truncate_partial_nodes().
1) An err is occurred in the first 'for' loop
When err is occurred with depth = 2, pages[0] is invalid, so this page doesn't
need to be put. There is no problem, however, when depth is 3, it doesn't put
the pages correctly where pages[0] is valid and pages[1] is invalid.
In this case, depth is set to 2 (ref to statemnt depth = i + 1), and then
'goto fail'.
In label 'fail', for (i = depth - 3; i >= 0; i--) cannot meet the condition
because i = -1, so pages[0] cann't be put.
2) An err happened in the second 'for' loop
Now we've got pages[0] with depth = 2, or we've got pages[0] and pages[1]
with depth = 3. When an err is detected, we need 'goto fail' to put such
the pages.
When depth is 2, in label 'fail', for (i = depth - 3; i >= 0; i--) cann't
meet the condition because i = -1, so pages[0] cann't be put.
When depth is 3, in label 'fail', for (i = depth - 3; i >= 0; i--) can
only put pages[0], pages[1] also cann't be put.
Note that 'depth' has been changed before first 'goto fail' (ref to statemnt
depth = i + 1), so passing this modified 'depth' to the tracepoint,
trace_f2fs_truncate_partial_nodes, is also incorrect.
Signed-off-by: Shifei Ge <shifei10.ge@samsung.com>
[Jaegeuk Kim: modify the description and fix one bug]
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-29 07:32:34 +00:00
|
|
|
for (i = offset[idx + 1]; i < NIDS_PER_BLOCK; i++) {
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
child_nid = get_nid(pages[idx], i, false);
|
|
|
|
if (!child_nid)
|
|
|
|
continue;
|
|
|
|
dn->nid = child_nid;
|
|
|
|
err = truncate_dnode(dn);
|
|
|
|
if (err < 0)
|
|
|
|
goto fail;
|
2016-01-07 21:23:12 +00:00
|
|
|
if (set_nid(pages[idx], i, 0, false))
|
|
|
|
dn->node_changed = true;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
f2fs: fix truncate_partial_nodes bug
The truncate_partial_nodes puts pages incorrectly in the following two cases.
Note that the value for argc 'depth' can only be 2 or 3.
Please see truncate_inode_blocks() and truncate_partial_nodes().
1) An err is occurred in the first 'for' loop
When err is occurred with depth = 2, pages[0] is invalid, so this page doesn't
need to be put. There is no problem, however, when depth is 3, it doesn't put
the pages correctly where pages[0] is valid and pages[1] is invalid.
In this case, depth is set to 2 (ref to statemnt depth = i + 1), and then
'goto fail'.
In label 'fail', for (i = depth - 3; i >= 0; i--) cannot meet the condition
because i = -1, so pages[0] cann't be put.
2) An err happened in the second 'for' loop
Now we've got pages[0] with depth = 2, or we've got pages[0] and pages[1]
with depth = 3. When an err is detected, we need 'goto fail' to put such
the pages.
When depth is 2, in label 'fail', for (i = depth - 3; i >= 0; i--) cann't
meet the condition because i = -1, so pages[0] cann't be put.
When depth is 3, in label 'fail', for (i = depth - 3; i >= 0; i--) can
only put pages[0], pages[1] also cann't be put.
Note that 'depth' has been changed before first 'goto fail' (ref to statemnt
depth = i + 1), so passing this modified 'depth' to the tracepoint,
trace_f2fs_truncate_partial_nodes, is also incorrect.
Signed-off-by: Shifei Ge <shifei10.ge@samsung.com>
[Jaegeuk Kim: modify the description and fix one bug]
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-29 07:32:34 +00:00
|
|
|
if (offset[idx + 1] == 0) {
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
dn->node_page = pages[idx];
|
|
|
|
dn->nid = nid[idx];
|
|
|
|
truncate_node(dn);
|
|
|
|
} else {
|
|
|
|
f2fs_put_page(pages[idx], 1);
|
|
|
|
}
|
|
|
|
offset[idx]++;
|
f2fs: fix truncate_partial_nodes bug
The truncate_partial_nodes puts pages incorrectly in the following two cases.
Note that the value for argc 'depth' can only be 2 or 3.
Please see truncate_inode_blocks() and truncate_partial_nodes().
1) An err is occurred in the first 'for' loop
When err is occurred with depth = 2, pages[0] is invalid, so this page doesn't
need to be put. There is no problem, however, when depth is 3, it doesn't put
the pages correctly where pages[0] is valid and pages[1] is invalid.
In this case, depth is set to 2 (ref to statemnt depth = i + 1), and then
'goto fail'.
In label 'fail', for (i = depth - 3; i >= 0; i--) cannot meet the condition
because i = -1, so pages[0] cann't be put.
2) An err happened in the second 'for' loop
Now we've got pages[0] with depth = 2, or we've got pages[0] and pages[1]
with depth = 3. When an err is detected, we need 'goto fail' to put such
the pages.
When depth is 2, in label 'fail', for (i = depth - 3; i >= 0; i--) cann't
meet the condition because i = -1, so pages[0] cann't be put.
When depth is 3, in label 'fail', for (i = depth - 3; i >= 0; i--) can
only put pages[0], pages[1] also cann't be put.
Note that 'depth' has been changed before first 'goto fail' (ref to statemnt
depth = i + 1), so passing this modified 'depth' to the tracepoint,
trace_f2fs_truncate_partial_nodes, is also incorrect.
Signed-off-by: Shifei Ge <shifei10.ge@samsung.com>
[Jaegeuk Kim: modify the description and fix one bug]
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-29 07:32:34 +00:00
|
|
|
offset[idx + 1] = 0;
|
|
|
|
idx--;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
fail:
|
f2fs: fix truncate_partial_nodes bug
The truncate_partial_nodes puts pages incorrectly in the following two cases.
Note that the value for argc 'depth' can only be 2 or 3.
Please see truncate_inode_blocks() and truncate_partial_nodes().
1) An err is occurred in the first 'for' loop
When err is occurred with depth = 2, pages[0] is invalid, so this page doesn't
need to be put. There is no problem, however, when depth is 3, it doesn't put
the pages correctly where pages[0] is valid and pages[1] is invalid.
In this case, depth is set to 2 (ref to statemnt depth = i + 1), and then
'goto fail'.
In label 'fail', for (i = depth - 3; i >= 0; i--) cannot meet the condition
because i = -1, so pages[0] cann't be put.
2) An err happened in the second 'for' loop
Now we've got pages[0] with depth = 2, or we've got pages[0] and pages[1]
with depth = 3. When an err is detected, we need 'goto fail' to put such
the pages.
When depth is 2, in label 'fail', for (i = depth - 3; i >= 0; i--) cann't
meet the condition because i = -1, so pages[0] cann't be put.
When depth is 3, in label 'fail', for (i = depth - 3; i >= 0; i--) can
only put pages[0], pages[1] also cann't be put.
Note that 'depth' has been changed before first 'goto fail' (ref to statemnt
depth = i + 1), so passing this modified 'depth' to the tracepoint,
trace_f2fs_truncate_partial_nodes, is also incorrect.
Signed-off-by: Shifei Ge <shifei10.ge@samsung.com>
[Jaegeuk Kim: modify the description and fix one bug]
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-10-29 07:32:34 +00:00
|
|
|
for (i = idx; i >= 0; i--)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
f2fs_put_page(pages[i], 1);
|
2013-04-19 16:28:52 +00:00
|
|
|
|
|
|
|
trace_f2fs_truncate_partial_nodes(dn->inode, nid, depth, err);
|
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2012-11-29 04:28:09 +00:00
|
|
|
/*
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
* All the block addresses of data and nodes should be nullified.
|
|
|
|
*/
|
|
|
|
int truncate_inode_blocks(struct inode *inode, pgoff_t from)
|
|
|
|
{
|
2014-09-02 22:31:18 +00:00
|
|
|
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
int err = 0, cont = 1;
|
|
|
|
int level, offset[4], noffset[4];
|
2013-02-11 22:28:55 +00:00
|
|
|
unsigned int nofs = 0;
|
2013-12-26 07:30:41 +00:00
|
|
|
struct f2fs_inode *ri;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
struct dnode_of_data dn;
|
|
|
|
struct page *page;
|
|
|
|
|
2013-04-19 16:28:52 +00:00
|
|
|
trace_f2fs_truncate_inode_blocks_enter(inode, from);
|
|
|
|
|
2016-01-26 07:39:35 +00:00
|
|
|
level = get_node_path(inode, from, offset, noffset);
|
2017-08-21 20:51:32 +00:00
|
|
|
if (level < 0)
|
|
|
|
return level;
|
2016-03-29 23:13:45 +00:00
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
page = get_node_page(sbi, inode->i_ino);
|
2013-04-19 16:28:52 +00:00
|
|
|
if (IS_ERR(page)) {
|
|
|
|
trace_f2fs_truncate_inode_blocks_exit(inode, PTR_ERR(page));
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return PTR_ERR(page);
|
2013-04-19 16:28:52 +00:00
|
|
|
}
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
set_new_dnode(&dn, inode, page, NULL, 0);
|
|
|
|
unlock_page(page);
|
|
|
|
|
2013-12-26 07:30:41 +00:00
|
|
|
ri = F2FS_INODE(page);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
switch (level) {
|
|
|
|
case 0:
|
|
|
|
case 1:
|
|
|
|
nofs = noffset[1];
|
|
|
|
break;
|
|
|
|
case 2:
|
|
|
|
nofs = noffset[1];
|
|
|
|
if (!offset[level - 1])
|
|
|
|
goto skip_partial;
|
2013-12-26 07:30:41 +00:00
|
|
|
err = truncate_partial_nodes(&dn, ri, offset, level);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (err < 0 && err != -ENOENT)
|
|
|
|
goto fail;
|
|
|
|
nofs += 1 + NIDS_PER_BLOCK;
|
|
|
|
break;
|
|
|
|
case 3:
|
|
|
|
nofs = 5 + 2 * NIDS_PER_BLOCK;
|
|
|
|
if (!offset[level - 1])
|
|
|
|
goto skip_partial;
|
2013-12-26 07:30:41 +00:00
|
|
|
err = truncate_partial_nodes(&dn, ri, offset, level);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (err < 0 && err != -ENOENT)
|
|
|
|
goto fail;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
BUG();
|
|
|
|
}
|
|
|
|
|
|
|
|
skip_partial:
|
|
|
|
while (cont) {
|
2013-12-26 07:30:41 +00:00
|
|
|
dn.nid = le32_to_cpu(ri->i_nid[offset[0] - NODE_DIR1_BLOCK]);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
switch (offset[0]) {
|
|
|
|
case NODE_DIR1_BLOCK:
|
|
|
|
case NODE_DIR2_BLOCK:
|
|
|
|
err = truncate_dnode(&dn);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case NODE_IND1_BLOCK:
|
|
|
|
case NODE_IND2_BLOCK:
|
|
|
|
err = truncate_nodes(&dn, nofs, offset[1], 2);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case NODE_DIND_BLOCK:
|
|
|
|
err = truncate_nodes(&dn, nofs, offset[1], 3);
|
|
|
|
cont = 0;
|
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
|
|
|
BUG();
|
|
|
|
}
|
|
|
|
if (err < 0 && err != -ENOENT)
|
|
|
|
goto fail;
|
|
|
|
if (offset[1] == 0 &&
|
2013-12-26 07:30:41 +00:00
|
|
|
ri->i_nid[offset[0] - NODE_DIR1_BLOCK]) {
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
lock_page(page);
|
2016-03-29 23:13:45 +00:00
|
|
|
BUG_ON(page->mapping != NODE_MAPPING(sbi));
|
2016-01-20 15:43:51 +00:00
|
|
|
f2fs_wait_on_page_writeback(page, NODE, true);
|
2013-12-26 07:30:41 +00:00
|
|
|
ri->i_nid[offset[0] - NODE_DIR1_BLOCK] = 0;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
set_page_dirty(page);
|
|
|
|
unlock_page(page);
|
|
|
|
}
|
|
|
|
offset[1] = 0;
|
|
|
|
offset[0]++;
|
|
|
|
nofs += err;
|
|
|
|
}
|
|
|
|
fail:
|
|
|
|
f2fs_put_page(page, 0);
|
2013-04-19 16:28:52 +00:00
|
|
|
trace_f2fs_truncate_inode_blocks_exit(inode, err);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return err > 0 ? 0 : err;
|
|
|
|
}
|
|
|
|
|
2017-10-19 18:48:57 +00:00
|
|
|
/* caller must lock inode page */
|
|
|
|
int truncate_xattr_node(struct inode *inode)
|
2013-08-14 11:40:06 +00:00
|
|
|
{
|
2014-09-02 22:31:18 +00:00
|
|
|
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
|
2013-08-14 11:40:06 +00:00
|
|
|
nid_t nid = F2FS_I(inode)->i_xattr_nid;
|
|
|
|
struct dnode_of_data dn;
|
|
|
|
struct page *npage;
|
|
|
|
|
|
|
|
if (!nid)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
npage = get_node_page(sbi, nid);
|
|
|
|
if (IS_ERR(npage))
|
|
|
|
return PTR_ERR(npage);
|
|
|
|
|
2016-05-20 16:52:20 +00:00
|
|
|
f2fs_i_xnid_write(inode, 0);
|
2013-08-14 12:57:27 +00:00
|
|
|
|
2017-10-19 18:48:57 +00:00
|
|
|
set_new_dnode(&dn, inode, NULL, npage, nid);
|
2013-08-14 11:40:06 +00:00
|
|
|
truncate_node(&dn);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
f2fs: introduce a new global lock scheme
In the previous version, f2fs uses global locks according to the usage types,
such as directory operations, block allocation, block write, and so on.
Reference the following lock types in f2fs.h.
enum lock_type {
RENAME, /* for renaming operations */
DENTRY_OPS, /* for directory operations */
DATA_WRITE, /* for data write */
DATA_NEW, /* for data allocation */
DATA_TRUNC, /* for data truncate */
NODE_NEW, /* for node allocation */
NODE_TRUNC, /* for node truncate */
NODE_WRITE, /* for node write */
NR_LOCK_TYPE,
};
In that case, we lose the performance under the multi-threading environment,
since every types of operations must be conducted one at a time.
In order to address the problem, let's share the locks globally with a mutex
array regardless of any types.
So, let users grab a mutex and perform their jobs in parallel as much as
possbile.
For this, I propose a new global lock scheme as follows.
0. Data structure
- f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS]
- f2fs_sb_info -> node_write
1. mutex_lock_op(sbi)
- try to get an avaiable lock from the array.
- returns the index of the gottern lock variable.
2. mutex_unlock_op(sbi, index of the lock)
- unlock the given index of the lock.
3. mutex_lock_all(sbi)
- grab all the locks in the array before the checkpoint.
4. mutex_unlock_all(sbi)
- release all the locks in the array after checkpoint.
5. block_operations()
- call mutex_lock_all()
- sync_dirty_dir_inodes()
- grab node_write
- sync_node_pages()
Note that,
the pairs of mutex_lock_op()/mutex_unlock_op() and
mutex_lock_all()/mutex_unlock_all() should be used together.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
|
|
|
/*
|
2013-12-21 10:02:14 +00:00
|
|
|
* Caller should grab and release a rwsem by calling f2fs_lock_op() and
|
|
|
|
* f2fs_unlock_op().
|
f2fs: introduce a new global lock scheme
In the previous version, f2fs uses global locks according to the usage types,
such as directory operations, block allocation, block write, and so on.
Reference the following lock types in f2fs.h.
enum lock_type {
RENAME, /* for renaming operations */
DENTRY_OPS, /* for directory operations */
DATA_WRITE, /* for data write */
DATA_NEW, /* for data allocation */
DATA_TRUNC, /* for data truncate */
NODE_NEW, /* for node allocation */
NODE_TRUNC, /* for node truncate */
NODE_WRITE, /* for node write */
NR_LOCK_TYPE,
};
In that case, we lose the performance under the multi-threading environment,
since every types of operations must be conducted one at a time.
In order to address the problem, let's share the locks globally with a mutex
array regardless of any types.
So, let users grab a mutex and perform their jobs in parallel as much as
possbile.
For this, I propose a new global lock scheme as follows.
0. Data structure
- f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS]
- f2fs_sb_info -> node_write
1. mutex_lock_op(sbi)
- try to get an avaiable lock from the array.
- returns the index of the gottern lock variable.
2. mutex_unlock_op(sbi, index of the lock)
- unlock the given index of the lock.
3. mutex_lock_all(sbi)
- grab all the locks in the array before the checkpoint.
4. mutex_unlock_all(sbi)
- release all the locks in the array after checkpoint.
5. block_operations()
- call mutex_lock_all()
- sync_dirty_dir_inodes()
- grab node_write
- sync_node_pages()
Note that,
the pairs of mutex_lock_op()/mutex_unlock_op() and
mutex_lock_all()/mutex_unlock_all() should be used together.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
|
|
|
*/
|
2015-08-24 09:40:45 +00:00
|
|
|
int remove_inode_page(struct inode *inode)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
|
|
|
struct dnode_of_data dn;
|
2015-08-24 09:40:45 +00:00
|
|
|
int err;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2014-08-25 21:45:59 +00:00
|
|
|
set_new_dnode(&dn, inode, NULL, NULL, inode->i_ino);
|
2015-08-24 09:40:45 +00:00
|
|
|
err = get_dnode_of_data(&dn, 0, LOOKUP_NODE);
|
|
|
|
if (err)
|
|
|
|
return err;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2017-10-19 18:48:57 +00:00
|
|
|
err = truncate_xattr_node(inode);
|
2015-08-24 09:40:45 +00:00
|
|
|
if (err) {
|
2014-08-25 21:45:59 +00:00
|
|
|
f2fs_put_dnode(&dn);
|
2015-08-24 09:40:45 +00:00
|
|
|
return err;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
2014-08-25 21:45:59 +00:00
|
|
|
|
|
|
|
/* remove potential inline_data blocks */
|
|
|
|
if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
|
|
|
|
S_ISLNK(inode->i_mode))
|
|
|
|
truncate_data_blocks_range(&dn, 1);
|
|
|
|
|
2014-08-06 14:22:50 +00:00
|
|
|
/* 0 is possible, after f2fs_new_inode() has failed */
|
2014-09-02 22:52:58 +00:00
|
|
|
f2fs_bug_on(F2FS_I_SB(inode),
|
2017-06-14 15:00:56 +00:00
|
|
|
inode->i_blocks != 0 && inode->i_blocks != 8);
|
2014-08-25 21:45:59 +00:00
|
|
|
|
|
|
|
/* will put inode & node pages */
|
2012-12-20 06:10:06 +00:00
|
|
|
truncate_node(&dn);
|
2015-08-24 09:40:45 +00:00
|
|
|
return 0;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2014-06-21 04:44:02 +00:00
|
|
|
struct page *new_inode_page(struct inode *inode)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
|
|
|
struct dnode_of_data dn;
|
|
|
|
|
|
|
|
/* allocate inode page for new inode */
|
|
|
|
set_new_dnode(&dn, inode, NULL, NULL, inode->i_ino);
|
2013-05-20 01:10:29 +00:00
|
|
|
|
|
|
|
/* caller should f2fs_put_page(page, 1); */
|
2017-07-17 11:16:11 +00:00
|
|
|
return new_node_page(&dn, 0);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2017-07-17 11:16:11 +00:00
|
|
|
struct page *new_node_page(struct dnode_of_data *dn, unsigned int ofs)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
2014-09-02 22:31:18 +00:00
|
|
|
struct f2fs_sb_info *sbi = F2FS_I_SB(dn->inode);
|
2017-02-14 01:02:44 +00:00
|
|
|
struct node_info new_ni;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
struct page *page;
|
|
|
|
int err;
|
|
|
|
|
2016-05-20 17:13:22 +00:00
|
|
|
if (unlikely(is_inode_flag_set(dn->inode, FI_NO_ALLOC)))
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return ERR_PTR(-EPERM);
|
|
|
|
|
2016-04-29 23:11:53 +00:00
|
|
|
page = f2fs_grab_cache_page(NODE_MAPPING(sbi), dn->nid, false);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (!page)
|
|
|
|
return ERR_PTR(-ENOMEM);
|
|
|
|
|
2017-07-08 16:13:07 +00:00
|
|
|
if (unlikely((err = inc_valid_node_count(sbi, dn->inode, !ofs))))
|
2013-08-12 07:00:46 +00:00
|
|
|
goto fail;
|
2017-07-08 16:13:07 +00:00
|
|
|
|
2017-02-14 01:02:44 +00:00
|
|
|
#ifdef CONFIG_F2FS_CHECK_FS
|
|
|
|
get_node_info(sbi, dn->nid, &new_ni);
|
|
|
|
f2fs_bug_on(sbi, new_ni.blk_addr != NULL_ADDR);
|
|
|
|
#endif
|
|
|
|
new_ni.nid = dn->nid;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
new_ni.ino = dn->inode->i_ino;
|
2017-02-14 01:02:44 +00:00
|
|
|
new_ni.blk_addr = NULL_ADDR;
|
|
|
|
new_ni.flag = 0;
|
|
|
|
new_ni.version = 0;
|
2014-03-20 12:52:53 +00:00
|
|
|
set_node_addr(sbi, &new_ni, NEW_ADDR, false);
|
2013-08-12 07:00:46 +00:00
|
|
|
|
2016-01-20 15:43:51 +00:00
|
|
|
f2fs_wait_on_page_writeback(page, NODE, true);
|
2013-08-12 07:00:46 +00:00
|
|
|
fill_node_footer(page, dn->nid, dn->inode->i_ino, ofs, true);
|
2018-03-09 06:24:22 +00:00
|
|
|
set_cold_node(page, S_ISDIR(dn->inode->i_mode));
|
2016-07-01 01:49:15 +00:00
|
|
|
if (!PageUptodate(page))
|
|
|
|
SetPageUptodate(page);
|
2016-01-07 21:23:12 +00:00
|
|
|
if (set_page_dirty(page))
|
|
|
|
dn->node_changed = true;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2014-03-17 08:35:06 +00:00
|
|
|
if (f2fs_has_xattr_block(ofs))
|
2016-05-20 16:52:20 +00:00
|
|
|
f2fs_i_xnid_write(dn->inode, dn->nid);
|
2013-08-12 07:04:53 +00:00
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (ofs == 0)
|
|
|
|
inc_valid_inode_count(sbi);
|
|
|
|
return page;
|
|
|
|
|
|
|
|
fail:
|
2012-12-20 06:10:06 +00:00
|
|
|
clear_node_page_dirty(page);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
f2fs_put_page(page, 1);
|
|
|
|
return ERR_PTR(err);
|
|
|
|
}
|
|
|
|
|
2013-03-31 03:47:20 +00:00
|
|
|
/*
|
|
|
|
* Caller should do after getting the following values.
|
|
|
|
* 0: f2fs_put_page(page, 0)
|
2015-07-15 20:08:21 +00:00
|
|
|
* LOCKED_PAGE or error: f2fs_put_page(page, 1)
|
2013-03-31 03:47:20 +00:00
|
|
|
*/
|
2016-06-05 19:31:55 +00:00
|
|
|
static int read_node_page(struct page *page, int op_flags)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
2014-09-02 22:31:18 +00:00
|
|
|
struct f2fs_sb_info *sbi = F2FS_P_SB(page);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
struct node_info ni;
|
2014-12-18 03:33:13 +00:00
|
|
|
struct f2fs_io_info fio = {
|
2015-04-23 21:38:15 +00:00
|
|
|
.sbi = sbi,
|
2014-12-18 03:33:13 +00:00
|
|
|
.type = NODE,
|
2016-06-05 19:31:55 +00:00
|
|
|
.op = REQ_OP_READ,
|
|
|
|
.op_flags = op_flags,
|
2015-04-23 21:38:15 +00:00
|
|
|
.page = page,
|
2015-04-23 19:04:33 +00:00
|
|
|
.encrypted_page = NULL,
|
2014-12-18 03:33:13 +00:00
|
|
|
};
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2016-07-01 02:04:16 +00:00
|
|
|
if (PageUptodate(page))
|
|
|
|
return LOCKED_PAGE;
|
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
get_node_info(sbi, page->index, &ni);
|
|
|
|
|
2013-12-06 06:00:58 +00:00
|
|
|
if (unlikely(ni.blk_addr == NULL_ADDR)) {
|
2015-02-26 03:25:01 +00:00
|
|
|
ClearPageUptodate(page);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return -ENOENT;
|
2013-03-08 12:29:23 +00:00
|
|
|
}
|
|
|
|
|
f2fs: trace old block address for CoWed page
This patch enables to trace old block address of CoWed page for better
debugging.
f2fs_submit_page_mbio: dev = (1,0), ino = 1, page_index = 0x1d4f0, oldaddr = 0xfe8ab, newaddr = 0xfee90 rw = WRITE_SYNC, type = NODE
f2fs_submit_page_mbio: dev = (1,0), ino = 1, page_index = 0x1d4f8, oldaddr = 0xfe8b0, newaddr = 0xfee91 rw = WRITE_SYNC, type = NODE
f2fs_submit_page_mbio: dev = (1,0), ino = 1, page_index = 0x1d4fa, oldaddr = 0xfe8ae, newaddr = 0xfee92 rw = WRITE_SYNC, type = NODE
f2fs_submit_page_mbio: dev = (1,0), ino = 134824, page_index = 0x96, oldaddr = 0xf049b, newaddr = 0x2bbe rw = WRITE, type = DATA
f2fs_submit_page_mbio: dev = (1,0), ino = 134824, page_index = 0x97, oldaddr = 0xf049c, newaddr = 0x2bbf rw = WRITE, type = DATA
f2fs_submit_page_mbio: dev = (1,0), ino = 134824, page_index = 0x98, oldaddr = 0xf049d, newaddr = 0x2bc0 rw = WRITE, type = DATA
f2fs_submit_page_mbio: dev = (1,0), ino = 135260, page_index = 0x47, oldaddr = 0xffffffff, newaddr = 0xf2631 rw = WRITE, type = DATA
f2fs_submit_page_mbio: dev = (1,0), ino = 135260, page_index = 0x48, oldaddr = 0xffffffff, newaddr = 0xf2632 rw = WRITE, type = DATA
f2fs_submit_page_mbio: dev = (1,0), ino = 135260, page_index = 0x49, oldaddr = 0xffffffff, newaddr = 0xf2633 rw = WRITE, type = DATA
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 10:36:38 +00:00
|
|
|
fio.new_blkaddr = fio.old_blkaddr = ni.blk_addr;
|
2015-04-23 21:38:15 +00:00
|
|
|
return f2fs_submit_page_bio(&fio);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2012-11-29 04:28:09 +00:00
|
|
|
/*
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
* Readahead a node page
|
|
|
|
*/
|
|
|
|
void ra_node_page(struct f2fs_sb_info *sbi, nid_t nid)
|
|
|
|
{
|
|
|
|
struct page *apage;
|
2013-03-31 03:47:20 +00:00
|
|
|
int err;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2016-01-08 12:13:37 +00:00
|
|
|
if (!nid)
|
|
|
|
return;
|
2018-04-24 05:02:31 +00:00
|
|
|
if (check_nid_range(sbi, nid))
|
|
|
|
return;
|
2016-01-08 12:13:37 +00:00
|
|
|
|
2016-02-29 06:29:51 +00:00
|
|
|
rcu_read_lock();
|
2018-04-10 23:36:56 +00:00
|
|
|
apage = radix_tree_lookup(&NODE_MAPPING(sbi)->i_pages, nid);
|
2016-02-29 06:29:51 +00:00
|
|
|
rcu_read_unlock();
|
|
|
|
if (apage)
|
2013-03-08 12:29:23 +00:00
|
|
|
return;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2016-04-29 23:11:53 +00:00
|
|
|
apage = f2fs_grab_cache_page(NODE_MAPPING(sbi), nid, false);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (!apage)
|
|
|
|
return;
|
|
|
|
|
2016-07-19 09:28:41 +00:00
|
|
|
err = read_node_page(apage, REQ_RAHEAD);
|
2015-07-15 20:08:21 +00:00
|
|
|
f2fs_put_page(apage, err ? 1 : 0);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2016-03-08 17:04:35 +00:00
|
|
|
static struct page *__get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid,
|
2016-01-05 08:52:46 +00:00
|
|
|
struct page *parent, int start)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
|
|
|
struct page *page;
|
2016-01-05 08:52:46 +00:00
|
|
|
int err;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
if (!nid)
|
|
|
|
return ERR_PTR(-ENOENT);
|
2018-04-24 05:02:31 +00:00
|
|
|
if (check_nid_range(sbi, nid))
|
|
|
|
return ERR_PTR(-EINVAL);
|
2013-04-26 02:55:17 +00:00
|
|
|
repeat:
|
2016-04-29 23:11:53 +00:00
|
|
|
page = f2fs_grab_cache_page(NODE_MAPPING(sbi), nid, false);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (!page)
|
|
|
|
return ERR_PTR(-ENOMEM);
|
|
|
|
|
2016-11-01 13:40:10 +00:00
|
|
|
err = read_node_page(page, 0);
|
2015-07-15 20:08:21 +00:00
|
|
|
if (err < 0) {
|
|
|
|
f2fs_put_page(page, 1);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return ERR_PTR(err);
|
2015-07-15 20:08:21 +00:00
|
|
|
} else if (err == LOCKED_PAGE) {
|
2017-06-07 03:17:35 +00:00
|
|
|
err = 0;
|
2013-03-31 03:47:20 +00:00
|
|
|
goto page_hit;
|
2015-07-15 20:08:21 +00:00
|
|
|
}
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2016-01-05 08:52:46 +00:00
|
|
|
if (parent)
|
2016-05-06 23:19:43 +00:00
|
|
|
ra_node_pages(parent, start + 1, MAX_RA_NODE);
|
f2fs: give a chance to merge IOs by IO scheduler
Previously, background GC submits many 4KB read requests to load victim blocks
and/or its (i)node blocks.
...
f2fs_gc : f2fs_readpage: ino = 1, page_index = 0xb61, blkaddr = 0x3b964ed
f2fs_gc : block_rq_complete: 8,16 R () 499854968 + 8 [0]
f2fs_gc : f2fs_readpage: ino = 1, page_index = 0xb6f, blkaddr = 0x3b964ee
f2fs_gc : block_rq_complete: 8,16 R () 499854976 + 8 [0]
f2fs_gc : f2fs_readpage: ino = 1, page_index = 0xb79, blkaddr = 0x3b964ef
f2fs_gc : block_rq_complete: 8,16 R () 499854984 + 8 [0]
...
However, by the fact that many IOs are sequential, we can give a chance to merge
the IOs by IO scheduler.
In order to do that, let's use blk_plug.
...
f2fs_gc : f2fs_iget: ino = 143
f2fs_gc : f2fs_readpage: ino = 143, page_index = 0x1c6, blkaddr = 0x2e6ee
f2fs_gc : f2fs_iget: ino = 143
f2fs_gc : f2fs_readpage: ino = 143, page_index = 0x1c7, blkaddr = 0x2e6ef
<idle> : block_rq_complete: 8,16 R () 1519616 + 8 [0]
<idle> : block_rq_complete: 8,16 R () 1519848 + 8 [0]
<idle> : block_rq_complete: 8,16 R () 1520432 + 96 [0]
<idle> : block_rq_complete: 8,16 R () 1520536 + 104 [0]
<idle> : block_rq_complete: 8,16 R () 1521008 + 112 [0]
<idle> : block_rq_complete: 8,16 R () 1521440 + 152 [0]
<idle> : block_rq_complete: 8,16 R () 1521688 + 144 [0]
<idle> : block_rq_complete: 8,16 R () 1522128 + 192 [0]
<idle> : block_rq_complete: 8,16 R () 1523256 + 328 [0]
...
Note that this issue should be addressed in checkpoint, and some readahead
flows too.
Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-04-24 04:19:56 +00:00
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
lock_page(page);
|
2016-01-05 08:52:46 +00:00
|
|
|
|
2014-01-21 09:51:16 +00:00
|
|
|
if (unlikely(page->mapping != NODE_MAPPING(sbi))) {
|
2013-04-26 02:55:17 +00:00
|
|
|
f2fs_put_page(page, 1);
|
|
|
|
goto repeat;
|
|
|
|
}
|
2016-07-03 14:05:12 +00:00
|
|
|
|
2017-06-07 03:17:35 +00:00
|
|
|
if (unlikely(!PageUptodate(page))) {
|
|
|
|
err = -EIO;
|
2016-07-03 14:05:12 +00:00
|
|
|
goto out_err;
|
2017-06-07 03:17:35 +00:00
|
|
|
}
|
2017-07-31 12:19:09 +00:00
|
|
|
|
|
|
|
if (!f2fs_inode_chksum_verify(sbi, page)) {
|
|
|
|
err = -EBADMSG;
|
|
|
|
goto out_err;
|
|
|
|
}
|
2013-02-02 14:51:51 +00:00
|
|
|
page_hit:
|
2016-05-26 11:40:29 +00:00
|
|
|
if(unlikely(nid != nid_of_node(page))) {
|
2017-06-07 03:17:35 +00:00
|
|
|
f2fs_msg(sbi->sb, KERN_WARNING, "inconsistent node block, "
|
|
|
|
"nid:%lu, node_footer[nid:%u,ino:%u,ofs:%u,cpver:%llu,blkaddr:%u]",
|
|
|
|
nid, nid_of_node(page), ino_of_node(page),
|
|
|
|
ofs_of_node(page), cpver_of_node(page),
|
|
|
|
next_blkaddr_of_node(page));
|
|
|
|
err = -EINVAL;
|
2016-05-26 11:40:29 +00:00
|
|
|
out_err:
|
2017-08-31 23:54:51 +00:00
|
|
|
ClearPageUptodate(page);
|
2016-05-26 11:40:29 +00:00
|
|
|
f2fs_put_page(page, 1);
|
2017-06-07 03:17:35 +00:00
|
|
|
return ERR_PTR(err);
|
2016-05-26 11:40:29 +00:00
|
|
|
}
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return page;
|
|
|
|
}
|
|
|
|
|
2016-01-05 08:52:46 +00:00
|
|
|
struct page *get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid)
|
|
|
|
{
|
|
|
|
return __get_node_page(sbi, nid, NULL, 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
struct page *get_node_page_ra(struct page *parent, int start)
|
|
|
|
{
|
|
|
|
struct f2fs_sb_info *sbi = F2FS_P_SB(parent);
|
|
|
|
nid_t nid = get_nid(parent, start, false);
|
|
|
|
|
|
|
|
return __get_node_page(sbi, nid, parent, start);
|
|
|
|
}
|
|
|
|
|
2016-01-25 13:57:05 +00:00
|
|
|
static void flush_inline_data(struct f2fs_sb_info *sbi, nid_t ino)
|
|
|
|
{
|
|
|
|
struct inode *inode;
|
|
|
|
struct page *page;
|
2016-05-20 16:11:09 +00:00
|
|
|
int ret;
|
2016-01-25 13:57:05 +00:00
|
|
|
|
|
|
|
/* should flush inline_data before evict_inode */
|
|
|
|
inode = ilookup(sbi->sb, ino);
|
|
|
|
if (!inode)
|
|
|
|
return;
|
|
|
|
|
2017-10-28 08:52:30 +00:00
|
|
|
page = f2fs_pagecache_get_page(inode->i_mapping, 0,
|
|
|
|
FGP_LOCK|FGP_NOWAIT, 0);
|
2016-01-25 13:57:05 +00:00
|
|
|
if (!page)
|
|
|
|
goto iput_out;
|
|
|
|
|
|
|
|
if (!PageUptodate(page))
|
|
|
|
goto page_out;
|
|
|
|
|
|
|
|
if (!PageDirty(page))
|
|
|
|
goto page_out;
|
|
|
|
|
|
|
|
if (!clear_page_dirty_for_io(page))
|
|
|
|
goto page_out;
|
|
|
|
|
2016-05-20 16:11:09 +00:00
|
|
|
ret = f2fs_write_inline_data(inode, page);
|
|
|
|
inode_dec_dirty_pages(inode);
|
2016-10-11 14:57:01 +00:00
|
|
|
remove_dirty_inode(inode);
|
2016-05-20 16:11:09 +00:00
|
|
|
if (ret)
|
2016-01-25 13:57:05 +00:00
|
|
|
set_page_dirty(page);
|
|
|
|
page_out:
|
2016-03-30 18:25:31 +00:00
|
|
|
f2fs_put_page(page, 1);
|
2016-01-25 13:57:05 +00:00
|
|
|
iput_out:
|
|
|
|
iput(inode);
|
|
|
|
}
|
|
|
|
|
2016-04-15 16:43:17 +00:00
|
|
|
static struct page *last_fsync_dnode(struct f2fs_sb_info *sbi, nid_t ino)
|
|
|
|
{
|
2017-11-16 01:34:51 +00:00
|
|
|
pgoff_t index;
|
2016-04-15 16:43:17 +00:00
|
|
|
struct pagevec pvec;
|
|
|
|
struct page *last_page = NULL;
|
2017-11-16 01:34:51 +00:00
|
|
|
int nr_pages;
|
2016-04-15 16:43:17 +00:00
|
|
|
|
2017-11-16 01:37:52 +00:00
|
|
|
pagevec_init(&pvec);
|
2016-04-15 16:43:17 +00:00
|
|
|
index = 0;
|
2017-11-16 01:34:51 +00:00
|
|
|
|
|
|
|
while ((nr_pages = pagevec_lookup_tag(&pvec, NODE_MAPPING(sbi), &index,
|
2017-11-16 01:35:19 +00:00
|
|
|
PAGECACHE_TAG_DIRTY))) {
|
2017-11-16 01:34:51 +00:00
|
|
|
int i;
|
2016-04-15 16:43:17 +00:00
|
|
|
|
|
|
|
for (i = 0; i < nr_pages; i++) {
|
|
|
|
struct page *page = pvec.pages[i];
|
|
|
|
|
|
|
|
if (unlikely(f2fs_cp_error(sbi))) {
|
|
|
|
f2fs_put_page(last_page, 0);
|
|
|
|
pagevec_release(&pvec);
|
|
|
|
return ERR_PTR(-EIO);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!IS_DNODE(page) || !is_cold_node(page))
|
|
|
|
continue;
|
|
|
|
if (ino_of_node(page) != ino)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
lock_page(page);
|
|
|
|
|
|
|
|
if (unlikely(page->mapping != NODE_MAPPING(sbi))) {
|
|
|
|
continue_unlock:
|
|
|
|
unlock_page(page);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
if (ino_of_node(page) != ino)
|
|
|
|
goto continue_unlock;
|
|
|
|
|
|
|
|
if (!PageDirty(page)) {
|
|
|
|
/* someone wrote it for us */
|
|
|
|
goto continue_unlock;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (last_page)
|
|
|
|
f2fs_put_page(last_page, 0);
|
|
|
|
|
|
|
|
get_page(page);
|
|
|
|
last_page = page;
|
|
|
|
unlock_page(page);
|
|
|
|
}
|
|
|
|
pagevec_release(&pvec);
|
|
|
|
cond_resched();
|
|
|
|
}
|
|
|
|
return last_page;
|
|
|
|
}
|
|
|
|
|
2017-02-04 01:44:04 +00:00
|
|
|
static int __write_node_page(struct page *page, bool atomic, bool *submitted,
|
2017-08-02 15:21:48 +00:00
|
|
|
struct writeback_control *wbc, bool do_balance,
|
|
|
|
enum iostat_type io_type)
|
2017-02-03 02:27:17 +00:00
|
|
|
{
|
|
|
|
struct f2fs_sb_info *sbi = F2FS_P_SB(page);
|
|
|
|
nid_t nid;
|
|
|
|
struct node_info ni;
|
|
|
|
struct f2fs_io_info fio = {
|
|
|
|
.sbi = sbi,
|
2017-09-29 05:59:38 +00:00
|
|
|
.ino = ino_of_node(page),
|
2017-02-03 02:27:17 +00:00
|
|
|
.type = NODE,
|
|
|
|
.op = REQ_OP_WRITE,
|
|
|
|
.op_flags = wbc_to_write_flags(wbc),
|
|
|
|
.page = page,
|
|
|
|
.encrypted_page = NULL,
|
2017-02-04 01:44:04 +00:00
|
|
|
.submitted = false,
|
2017-08-02 15:21:48 +00:00
|
|
|
.io_type = io_type,
|
2018-01-09 11:33:39 +00:00
|
|
|
.io_wbc = wbc,
|
2017-02-03 02:27:17 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
trace_f2fs_writepage(page, NODE);
|
|
|
|
|
2018-01-18 09:29:10 +00:00
|
|
|
if (unlikely(f2fs_cp_error(sbi))) {
|
|
|
|
dec_page_count(sbi, F2FS_DIRTY_NODES);
|
|
|
|
unlock_page(page);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2017-02-03 02:27:17 +00:00
|
|
|
if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
|
|
|
|
goto redirty_out;
|
|
|
|
|
|
|
|
/* get old block addr of this node page */
|
|
|
|
nid = nid_of_node(page);
|
|
|
|
f2fs_bug_on(sbi, page->index != nid);
|
|
|
|
|
|
|
|
if (wbc->for_reclaim) {
|
|
|
|
if (!down_read_trylock(&sbi->node_write))
|
|
|
|
goto redirty_out;
|
|
|
|
} else {
|
|
|
|
down_read(&sbi->node_write);
|
|
|
|
}
|
|
|
|
|
|
|
|
get_node_info(sbi, nid, &ni);
|
|
|
|
|
|
|
|
/* This page is already truncated */
|
|
|
|
if (unlikely(ni.blk_addr == NULL_ADDR)) {
|
|
|
|
ClearPageUptodate(page);
|
|
|
|
dec_page_count(sbi, F2FS_DIRTY_NODES);
|
|
|
|
up_read(&sbi->node_write);
|
|
|
|
unlock_page(page);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2017-02-03 02:18:06 +00:00
|
|
|
if (atomic && !test_opt(sbi, NOBARRIER))
|
|
|
|
fio.op_flags |= REQ_PREFLUSH | REQ_FUA;
|
|
|
|
|
2017-02-03 02:27:17 +00:00
|
|
|
set_page_writeback(page);
|
2018-04-12 06:09:04 +00:00
|
|
|
ClearPageError(page);
|
2017-02-03 02:27:17 +00:00
|
|
|
fio.old_blkaddr = ni.blk_addr;
|
|
|
|
write_node_page(nid, &fio);
|
|
|
|
set_node_addr(sbi, &ni, fio.new_blkaddr, is_fsync_dnode(page));
|
|
|
|
dec_page_count(sbi, F2FS_DIRTY_NODES);
|
|
|
|
up_read(&sbi->node_write);
|
|
|
|
|
2017-02-04 01:44:04 +00:00
|
|
|
if (wbc->for_reclaim) {
|
2017-05-10 18:28:38 +00:00
|
|
|
f2fs_submit_merged_write_cond(sbi, page->mapping->host, 0,
|
|
|
|
page->index, NODE);
|
2017-02-04 01:44:04 +00:00
|
|
|
submitted = NULL;
|
|
|
|
}
|
2017-02-03 02:27:17 +00:00
|
|
|
|
|
|
|
unlock_page(page);
|
|
|
|
|
2017-02-04 01:44:04 +00:00
|
|
|
if (unlikely(f2fs_cp_error(sbi))) {
|
2017-05-10 18:28:38 +00:00
|
|
|
f2fs_submit_merged_write(sbi, NODE);
|
2017-02-04 01:44:04 +00:00
|
|
|
submitted = NULL;
|
|
|
|
}
|
|
|
|
if (submitted)
|
|
|
|
*submitted = fio.submitted;
|
2017-02-03 02:27:17 +00:00
|
|
|
|
2017-07-27 12:11:00 +00:00
|
|
|
if (do_balance)
|
|
|
|
f2fs_balance_fs(sbi, false);
|
2017-02-03 02:27:17 +00:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
redirty_out:
|
|
|
|
redirty_page_for_writepage(wbc, page);
|
|
|
|
return AOP_WRITEPAGE_ACTIVATE;
|
|
|
|
}
|
|
|
|
|
2017-10-30 06:18:55 +00:00
|
|
|
void move_node_page(struct page *node_page, int gc_type)
|
|
|
|
{
|
|
|
|
if (gc_type == FG_GC) {
|
|
|
|
struct writeback_control wbc = {
|
|
|
|
.sync_mode = WB_SYNC_ALL,
|
|
|
|
.nr_to_write = 1,
|
|
|
|
.for_reclaim = 0,
|
|
|
|
};
|
|
|
|
|
|
|
|
set_page_dirty(node_page);
|
|
|
|
f2fs_wait_on_page_writeback(node_page, NODE, true);
|
|
|
|
|
|
|
|
f2fs_bug_on(F2FS_P_SB(node_page), PageWriteback(node_page));
|
|
|
|
if (!clear_page_dirty_for_io(node_page))
|
|
|
|
goto out_page;
|
|
|
|
|
|
|
|
if (__write_node_page(node_page, false, NULL,
|
|
|
|
&wbc, false, FS_GC_NODE_IO))
|
|
|
|
unlock_page(node_page);
|
|
|
|
goto release_page;
|
|
|
|
} else {
|
|
|
|
/* set page dirty and write it */
|
|
|
|
if (!PageWriteback(node_page))
|
|
|
|
set_page_dirty(node_page);
|
|
|
|
}
|
|
|
|
out_page:
|
|
|
|
unlock_page(node_page);
|
|
|
|
release_page:
|
|
|
|
f2fs_put_page(node_page, 0);
|
|
|
|
}
|
|
|
|
|
2017-02-03 02:27:17 +00:00
|
|
|
static int f2fs_write_node_page(struct page *page,
|
|
|
|
struct writeback_control *wbc)
|
|
|
|
{
|
2017-08-02 15:21:48 +00:00
|
|
|
return __write_node_page(page, false, NULL, wbc, false, FS_NODE_IO);
|
2017-02-03 02:27:17 +00:00
|
|
|
}
|
|
|
|
|
2016-05-21 03:42:37 +00:00
|
|
|
int fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
|
2016-04-15 16:43:17 +00:00
|
|
|
struct writeback_control *wbc, bool atomic)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
2017-11-16 01:34:51 +00:00
|
|
|
pgoff_t index;
|
2017-02-02 00:51:22 +00:00
|
|
|
pgoff_t last_idx = ULONG_MAX;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
struct pagevec pvec;
|
2016-04-15 16:25:04 +00:00
|
|
|
int ret = 0;
|
2016-04-15 16:43:17 +00:00
|
|
|
struct page *last_page = NULL;
|
|
|
|
bool marked = false;
|
2016-05-21 03:42:37 +00:00
|
|
|
nid_t ino = inode->i_ino;
|
2017-11-16 01:34:51 +00:00
|
|
|
int nr_pages;
|
2016-04-13 23:24:44 +00:00
|
|
|
|
2016-04-15 16:43:17 +00:00
|
|
|
if (atomic) {
|
|
|
|
last_page = last_fsync_dnode(sbi, ino);
|
|
|
|
if (IS_ERR_OR_NULL(last_page))
|
|
|
|
return PTR_ERR_OR_ZERO(last_page);
|
|
|
|
}
|
|
|
|
retry:
|
2017-11-16 01:37:52 +00:00
|
|
|
pagevec_init(&pvec);
|
2016-04-13 23:24:44 +00:00
|
|
|
index = 0;
|
2017-11-16 01:34:51 +00:00
|
|
|
|
|
|
|
while ((nr_pages = pagevec_lookup_tag(&pvec, NODE_MAPPING(sbi), &index,
|
2017-11-16 01:35:19 +00:00
|
|
|
PAGECACHE_TAG_DIRTY))) {
|
2017-11-16 01:34:51 +00:00
|
|
|
int i;
|
2016-04-13 23:24:44 +00:00
|
|
|
|
|
|
|
for (i = 0; i < nr_pages; i++) {
|
|
|
|
struct page *page = pvec.pages[i];
|
2017-02-04 01:44:04 +00:00
|
|
|
bool submitted = false;
|
2016-04-13 23:24:44 +00:00
|
|
|
|
|
|
|
if (unlikely(f2fs_cp_error(sbi))) {
|
2016-04-15 16:43:17 +00:00
|
|
|
f2fs_put_page(last_page, 0);
|
2016-04-13 23:24:44 +00:00
|
|
|
pagevec_release(&pvec);
|
2016-10-11 14:57:06 +00:00
|
|
|
ret = -EIO;
|
|
|
|
goto out;
|
2016-04-13 23:24:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if (!IS_DNODE(page) || !is_cold_node(page))
|
|
|
|
continue;
|
|
|
|
if (ino_of_node(page) != ino)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
lock_page(page);
|
|
|
|
|
|
|
|
if (unlikely(page->mapping != NODE_MAPPING(sbi))) {
|
|
|
|
continue_unlock:
|
|
|
|
unlock_page(page);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
if (ino_of_node(page) != ino)
|
|
|
|
goto continue_unlock;
|
|
|
|
|
2016-04-15 16:43:17 +00:00
|
|
|
if (!PageDirty(page) && page != last_page) {
|
2016-04-13 23:24:44 +00:00
|
|
|
/* someone wrote it for us */
|
|
|
|
goto continue_unlock;
|
|
|
|
}
|
|
|
|
|
|
|
|
f2fs_wait_on_page_writeback(page, NODE, true);
|
|
|
|
BUG_ON(PageWriteback(page));
|
|
|
|
|
2017-04-12 19:02:00 +00:00
|
|
|
set_fsync_mark(page, 0);
|
|
|
|
set_dentry_mark(page, 0);
|
|
|
|
|
2016-04-15 16:43:17 +00:00
|
|
|
if (!atomic || page == last_page) {
|
|
|
|
set_fsync_mark(page, 1);
|
2016-05-21 03:42:37 +00:00
|
|
|
if (IS_INODE(page)) {
|
|
|
|
if (is_inode_flag_set(inode,
|
|
|
|
FI_DIRTY_INODE))
|
|
|
|
update_inode(inode, page);
|
2016-04-15 16:43:17 +00:00
|
|
|
set_dentry_mark(page,
|
2016-04-13 23:24:44 +00:00
|
|
|
need_dentry_mark(sbi, ino));
|
2016-05-21 03:42:37 +00:00
|
|
|
}
|
2016-04-15 16:43:17 +00:00
|
|
|
/* may be written by other thread */
|
|
|
|
if (!PageDirty(page))
|
|
|
|
set_page_dirty(page);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!clear_page_dirty_for_io(page))
|
|
|
|
goto continue_unlock;
|
2016-04-13 23:24:44 +00:00
|
|
|
|
2017-02-03 02:18:06 +00:00
|
|
|
ret = __write_node_page(page, atomic &&
|
2017-02-04 01:44:04 +00:00
|
|
|
page == last_page,
|
2017-08-02 15:21:48 +00:00
|
|
|
&submitted, wbc, true,
|
|
|
|
FS_NODE_IO);
|
2016-04-15 16:25:04 +00:00
|
|
|
if (ret) {
|
2016-04-13 23:24:44 +00:00
|
|
|
unlock_page(page);
|
2016-04-15 16:43:17 +00:00
|
|
|
f2fs_put_page(last_page, 0);
|
|
|
|
break;
|
2017-02-04 01:44:04 +00:00
|
|
|
} else if (submitted) {
|
2017-02-02 00:51:22 +00:00
|
|
|
last_idx = page->index;
|
2016-04-15 16:43:17 +00:00
|
|
|
}
|
2016-09-29 10:50:10 +00:00
|
|
|
|
2016-04-15 16:43:17 +00:00
|
|
|
if (page == last_page) {
|
|
|
|
f2fs_put_page(page, 0);
|
|
|
|
marked = true;
|
2016-04-13 23:24:44 +00:00
|
|
|
break;
|
2016-04-15 16:25:04 +00:00
|
|
|
}
|
2016-04-13 23:24:44 +00:00
|
|
|
}
|
|
|
|
pagevec_release(&pvec);
|
|
|
|
cond_resched();
|
|
|
|
|
2016-04-15 16:43:17 +00:00
|
|
|
if (ret || marked)
|
2016-04-13 23:24:44 +00:00
|
|
|
break;
|
|
|
|
}
|
2016-04-15 16:43:17 +00:00
|
|
|
if (!ret && atomic && !marked) {
|
|
|
|
f2fs_msg(sbi->sb, KERN_DEBUG,
|
|
|
|
"Retry to write fsync mark: ino=%u, idx=%lx",
|
|
|
|
ino, last_page->index);
|
|
|
|
lock_page(last_page);
|
2016-11-16 09:26:24 +00:00
|
|
|
f2fs_wait_on_page_writeback(last_page, NODE, true);
|
2016-04-15 16:43:17 +00:00
|
|
|
set_page_dirty(last_page);
|
|
|
|
unlock_page(last_page);
|
|
|
|
goto retry;
|
|
|
|
}
|
2016-10-11 14:57:06 +00:00
|
|
|
out:
|
2017-02-02 00:51:22 +00:00
|
|
|
if (last_idx != ULONG_MAX)
|
2017-05-10 18:28:38 +00:00
|
|
|
f2fs_submit_merged_write_cond(sbi, NULL, ino, last_idx, NODE);
|
2016-04-15 16:25:04 +00:00
|
|
|
return ret ? -EIO: 0;
|
2016-04-13 23:24:44 +00:00
|
|
|
}
|
|
|
|
|
2017-07-27 12:11:00 +00:00
|
|
|
int sync_node_pages(struct f2fs_sb_info *sbi, struct writeback_control *wbc,
|
2017-08-02 15:21:48 +00:00
|
|
|
bool do_balance, enum iostat_type io_type)
|
2016-04-13 23:24:44 +00:00
|
|
|
{
|
2017-11-16 01:34:51 +00:00
|
|
|
pgoff_t index;
|
2016-04-13 23:24:44 +00:00
|
|
|
struct pagevec pvec;
|
|
|
|
int step = 0;
|
2016-03-11 23:33:22 +00:00
|
|
|
int nwritten = 0;
|
2016-09-29 10:50:10 +00:00
|
|
|
int ret = 0;
|
2017-11-16 01:34:51 +00:00
|
|
|
int nr_pages;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2017-11-16 01:37:52 +00:00
|
|
|
pagevec_init(&pvec);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
next_step:
|
|
|
|
index = 0;
|
2017-11-16 01:34:51 +00:00
|
|
|
|
|
|
|
while ((nr_pages = pagevec_lookup_tag(&pvec, NODE_MAPPING(sbi), &index,
|
2017-11-16 01:35:19 +00:00
|
|
|
PAGECACHE_TAG_DIRTY))) {
|
2017-11-16 01:34:51 +00:00
|
|
|
int i;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
for (i = 0; i < nr_pages; i++) {
|
|
|
|
struct page *page = pvec.pages[i];
|
2017-02-04 01:44:04 +00:00
|
|
|
bool submitted = false;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* flushing sequence with step:
|
|
|
|
* 0. indirect nodes
|
|
|
|
* 1. dentry dnodes
|
|
|
|
* 2. file dnodes
|
|
|
|
*/
|
|
|
|
if (step == 0 && IS_DNODE(page))
|
|
|
|
continue;
|
|
|
|
if (step == 1 && (!IS_DNODE(page) ||
|
|
|
|
is_cold_node(page)))
|
|
|
|
continue;
|
|
|
|
if (step == 2 && (!IS_DNODE(page) ||
|
|
|
|
!is_cold_node(page)))
|
|
|
|
continue;
|
2016-02-22 10:35:46 +00:00
|
|
|
lock_node:
|
2016-04-13 23:24:44 +00:00
|
|
|
if (!trylock_page(page))
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
continue;
|
|
|
|
|
2014-01-21 09:51:16 +00:00
|
|
|
if (unlikely(page->mapping != NODE_MAPPING(sbi))) {
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
continue_unlock:
|
|
|
|
unlock_page(page);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!PageDirty(page)) {
|
|
|
|
/* someone wrote it for us */
|
|
|
|
goto continue_unlock;
|
|
|
|
}
|
|
|
|
|
2016-01-25 13:57:05 +00:00
|
|
|
/* flush inline_data */
|
2016-04-13 23:24:44 +00:00
|
|
|
if (is_inline_node(page)) {
|
2016-01-25 13:57:05 +00:00
|
|
|
clear_inline_node(page);
|
|
|
|
unlock_page(page);
|
|
|
|
flush_inline_data(sbi, ino_of_node(page));
|
2016-02-22 10:35:46 +00:00
|
|
|
goto lock_node;
|
2016-01-25 13:57:05 +00:00
|
|
|
}
|
|
|
|
|
2016-01-28 19:48:52 +00:00
|
|
|
f2fs_wait_on_page_writeback(page, NODE, true);
|
|
|
|
|
|
|
|
BUG_ON(PageWriteback(page));
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (!clear_page_dirty_for_io(page))
|
|
|
|
goto continue_unlock;
|
|
|
|
|
2016-04-13 23:24:44 +00:00
|
|
|
set_fsync_mark(page, 0);
|
|
|
|
set_dentry_mark(page, 0);
|
2014-08-12 01:18:36 +00:00
|
|
|
|
2017-07-27 12:11:00 +00:00
|
|
|
ret = __write_node_page(page, false, &submitted,
|
2017-08-02 15:21:48 +00:00
|
|
|
wbc, do_balance, io_type);
|
2017-02-04 01:44:04 +00:00
|
|
|
if (ret)
|
2014-08-12 01:18:36 +00:00
|
|
|
unlock_page(page);
|
2017-02-04 01:44:04 +00:00
|
|
|
else if (submitted)
|
2016-09-29 10:50:10 +00:00
|
|
|
nwritten++;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
if (--wbc->nr_to_write == 0)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
pagevec_release(&pvec);
|
|
|
|
cond_resched();
|
|
|
|
|
|
|
|
if (wbc->nr_to_write == 0) {
|
|
|
|
step = 2;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (step < 2) {
|
|
|
|
step++;
|
|
|
|
goto next_step;
|
|
|
|
}
|
2018-01-18 09:29:10 +00:00
|
|
|
|
2016-09-29 10:50:10 +00:00
|
|
|
if (nwritten)
|
2017-05-10 18:28:38 +00:00
|
|
|
f2fs_submit_merged_write(sbi, NODE);
|
2018-01-18 09:29:10 +00:00
|
|
|
|
|
|
|
if (unlikely(f2fs_cp_error(sbi)))
|
|
|
|
return -EIO;
|
2016-09-29 10:50:10 +00:00
|
|
|
return ret;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2013-10-31 05:57:01 +00:00
|
|
|
int wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, nid_t ino)
|
|
|
|
{
|
2017-11-16 01:34:51 +00:00
|
|
|
pgoff_t index = 0;
|
2013-10-31 05:57:01 +00:00
|
|
|
struct pagevec pvec;
|
2016-09-16 10:44:21 +00:00
|
|
|
int ret2, ret = 0;
|
2017-11-16 01:34:51 +00:00
|
|
|
int nr_pages;
|
2013-10-31 05:57:01 +00:00
|
|
|
|
2017-11-16 01:37:52 +00:00
|
|
|
pagevec_init(&pvec);
|
2014-01-21 09:51:16 +00:00
|
|
|
|
2017-11-16 01:34:51 +00:00
|
|
|
while ((nr_pages = pagevec_lookup_tag(&pvec, NODE_MAPPING(sbi), &index,
|
2017-11-16 01:35:19 +00:00
|
|
|
PAGECACHE_TAG_WRITEBACK))) {
|
2017-11-16 01:34:51 +00:00
|
|
|
int i;
|
2013-10-31 05:57:01 +00:00
|
|
|
|
|
|
|
for (i = 0; i < nr_pages; i++) {
|
|
|
|
struct page *page = pvec.pages[i];
|
|
|
|
|
2013-11-04 02:28:33 +00:00
|
|
|
if (ino && ino_of_node(page) == ino) {
|
2016-01-20 15:43:51 +00:00
|
|
|
f2fs_wait_on_page_writeback(page, NODE, true);
|
2013-11-04 02:28:33 +00:00
|
|
|
if (TestClearPageError(page))
|
|
|
|
ret = -EIO;
|
|
|
|
}
|
2013-10-31 05:57:01 +00:00
|
|
|
}
|
|
|
|
pagevec_release(&pvec);
|
|
|
|
cond_resched();
|
|
|
|
}
|
|
|
|
|
2016-09-16 10:44:21 +00:00
|
|
|
ret2 = filemap_check_errors(NODE_MAPPING(sbi));
|
2013-10-31 05:57:01 +00:00
|
|
|
if (!ret)
|
|
|
|
ret = ret2;
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
static int f2fs_write_node_pages(struct address_space *mapping,
|
|
|
|
struct writeback_control *wbc)
|
|
|
|
{
|
2014-09-02 22:31:18 +00:00
|
|
|
struct f2fs_sb_info *sbi = F2FS_M_SB(mapping);
|
2016-07-14 02:33:19 +00:00
|
|
|
struct blk_plug plug;
|
2014-03-18 04:47:11 +00:00
|
|
|
long diff;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2017-06-29 15:20:45 +00:00
|
|
|
if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
|
|
|
|
goto skip_write;
|
|
|
|
|
2013-10-24 05:19:18 +00:00
|
|
|
/* balancing f2fs's metadata in background */
|
|
|
|
f2fs_balance_fs_bg(sbi);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2013-01-18 05:54:13 +00:00
|
|
|
/* collect a number of dirty node pages and write together */
|
2014-03-18 03:40:49 +00:00
|
|
|
if (get_pages(sbi, F2FS_DIRTY_NODES) < nr_pages_to_skip(sbi, NODE))
|
2014-03-18 04:43:05 +00:00
|
|
|
goto skip_write;
|
2013-01-18 05:54:13 +00:00
|
|
|
|
2016-02-04 08:14:00 +00:00
|
|
|
trace_f2fs_writepages(mapping->host, wbc, NODE);
|
|
|
|
|
2014-03-18 04:47:11 +00:00
|
|
|
diff = nr_pages_to_write(sbi, NODE, wbc);
|
2014-01-08 01:09:51 +00:00
|
|
|
wbc->sync_mode = WB_SYNC_NONE;
|
2016-07-14 02:33:19 +00:00
|
|
|
blk_start_plug(&plug);
|
2017-08-02 15:21:48 +00:00
|
|
|
sync_node_pages(sbi, wbc, true, FS_NODE_IO);
|
2016-07-14 02:33:19 +00:00
|
|
|
blk_finish_plug(&plug);
|
2014-03-18 04:47:11 +00:00
|
|
|
wbc->nr_to_write = max((long)0, wbc->nr_to_write - diff);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return 0;
|
2014-03-18 04:43:05 +00:00
|
|
|
|
|
|
|
skip_write:
|
|
|
|
wbc->pages_skipped += get_pages(sbi, F2FS_DIRTY_NODES);
|
2016-02-04 08:14:00 +00:00
|
|
|
trace_f2fs_writepages(mapping->host, wbc, NODE);
|
2014-03-18 04:43:05 +00:00
|
|
|
return 0;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int f2fs_set_node_page_dirty(struct page *page)
|
|
|
|
{
|
2013-10-24 08:53:29 +00:00
|
|
|
trace_f2fs_set_page_dirty(page, NODE);
|
|
|
|
|
2016-07-01 01:49:15 +00:00
|
|
|
if (!PageUptodate(page))
|
|
|
|
SetPageUptodate(page);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (!PageDirty(page)) {
|
2018-04-21 02:29:52 +00:00
|
|
|
__set_page_dirty_nobuffers(page);
|
2014-09-02 22:31:18 +00:00
|
|
|
inc_page_count(F2FS_P_SB(page), F2FS_DIRTY_NODES);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
SetPagePrivate(page);
|
2014-12-18 03:58:58 +00:00
|
|
|
f2fs_trace_pid(page);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2012-11-29 04:28:09 +00:00
|
|
|
/*
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
* Structure of the f2fs node operations
|
|
|
|
*/
|
|
|
|
const struct address_space_operations f2fs_node_aops = {
|
|
|
|
.writepage = f2fs_write_node_page,
|
|
|
|
.writepages = f2fs_write_node_pages,
|
|
|
|
.set_page_dirty = f2fs_set_node_page_dirty,
|
2015-02-05 09:44:29 +00:00
|
|
|
.invalidatepage = f2fs_invalidate_page,
|
|
|
|
.releasepage = f2fs_release_page,
|
2016-09-19 21:03:27 +00:00
|
|
|
#ifdef CONFIG_MIGRATION
|
|
|
|
.migratepage = f2fs_migrate_page,
|
|
|
|
#endif
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
};
|
|
|
|
|
2014-02-21 05:29:35 +00:00
|
|
|
static struct free_nid *__lookup_free_nid_list(struct f2fs_nm_info *nm_i,
|
|
|
|
nid_t n)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
2014-02-21 05:29:35 +00:00
|
|
|
return radix_tree_lookup(&nm_i->free_nid_root, n);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2017-09-29 05:59:35 +00:00
|
|
|
static int __insert_free_nid(struct f2fs_sb_info *sbi,
|
2017-10-28 11:03:37 +00:00
|
|
|
struct free_nid *i, enum nid_state state)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
|
2017-10-28 11:03:37 +00:00
|
|
|
int err = radix_tree_insert(&nm_i->free_nid_root, i->nid, i);
|
|
|
|
if (err)
|
|
|
|
return err;
|
2016-10-12 17:09:59 +00:00
|
|
|
|
2017-09-29 05:59:35 +00:00
|
|
|
f2fs_bug_on(sbi, state != i->state);
|
|
|
|
nm_i->nid_cnt[state]++;
|
|
|
|
if (state == FREE_NID)
|
|
|
|
list_add_tail(&i->list, &nm_i->free_nid_list);
|
2016-10-12 17:09:59 +00:00
|
|
|
return 0;
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
}
|
|
|
|
|
2017-09-29 05:59:35 +00:00
|
|
|
static void __remove_free_nid(struct f2fs_sb_info *sbi,
|
2017-10-28 11:03:37 +00:00
|
|
|
struct free_nid *i, enum nid_state state)
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
|
2017-09-29 05:59:35 +00:00
|
|
|
f2fs_bug_on(sbi, state != i->state);
|
|
|
|
nm_i->nid_cnt[state]--;
|
|
|
|
if (state == FREE_NID)
|
|
|
|
list_del(&i->list);
|
2017-10-28 11:03:37 +00:00
|
|
|
radix_tree_delete(&nm_i->free_nid_root, i->nid);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void __move_free_nid(struct f2fs_sb_info *sbi, struct free_nid *i,
|
|
|
|
enum nid_state org_state, enum nid_state dst_state)
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
|
2017-10-28 11:03:37 +00:00
|
|
|
f2fs_bug_on(sbi, org_state != i->state);
|
|
|
|
i->state = dst_state;
|
|
|
|
nm_i->nid_cnt[org_state]--;
|
|
|
|
nm_i->nid_cnt[dst_state]++;
|
|
|
|
|
|
|
|
switch (dst_state) {
|
|
|
|
case PREALLOC_NID:
|
|
|
|
list_del(&i->list);
|
|
|
|
break;
|
|
|
|
case FREE_NID:
|
|
|
|
list_add_tail(&i->list, &nm_i->free_nid_list);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
BUG_ON(1);
|
|
|
|
}
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2017-11-22 08:07:23 +00:00
|
|
|
static void update_free_nid_bitmap(struct f2fs_sb_info *sbi, nid_t nid,
|
|
|
|
bool set, bool build)
|
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
unsigned int nat_ofs = NAT_BLOCK_OFFSET(nid);
|
|
|
|
unsigned int nid_ofs = nid - START_NID(nid);
|
|
|
|
|
|
|
|
if (!test_bit_le(nat_ofs, nm_i->nat_block_bitmap))
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (set) {
|
|
|
|
if (test_bit_le(nid_ofs, nm_i->free_nid_bitmap[nat_ofs]))
|
|
|
|
return;
|
|
|
|
__set_bit_le(nid_ofs, nm_i->free_nid_bitmap[nat_ofs]);
|
|
|
|
nm_i->free_nid_count[nat_ofs]++;
|
|
|
|
} else {
|
|
|
|
if (!test_bit_le(nid_ofs, nm_i->free_nid_bitmap[nat_ofs]))
|
|
|
|
return;
|
|
|
|
__clear_bit_le(nid_ofs, nm_i->free_nid_bitmap[nat_ofs]);
|
|
|
|
if (!build)
|
|
|
|
nm_i->free_nid_count[nat_ofs]--;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
/* return if the nid is recognized as free */
|
2017-11-22 08:07:23 +00:00
|
|
|
static bool add_free_nid(struct f2fs_sb_info *sbi,
|
|
|
|
nid_t nid, bool build, bool update)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
2014-04-16 01:47:06 +00:00
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
2017-03-22 06:45:05 +00:00
|
|
|
struct free_nid *i, *e;
|
2013-05-07 11:47:40 +00:00
|
|
|
struct nat_entry *ne;
|
2017-03-22 06:45:05 +00:00
|
|
|
int err = -EINVAL;
|
|
|
|
bool ret = false;
|
2013-04-25 04:21:12 +00:00
|
|
|
|
|
|
|
/* 0 nid should not be used */
|
2013-12-05 09:15:22 +00:00
|
|
|
if (unlikely(nid == 0))
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
return false;
|
2013-05-07 11:47:40 +00:00
|
|
|
|
2013-10-22 06:52:26 +00:00
|
|
|
i = f2fs_kmem_cache_alloc(free_nid_slab, GFP_NOFS);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
i->nid = nid;
|
2017-09-29 05:59:35 +00:00
|
|
|
i->state = FREE_NID;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2017-11-22 08:07:23 +00:00
|
|
|
radix_tree_preload(GFP_NOFS | __GFP_NOFAIL);
|
2014-12-04 04:47:26 +00:00
|
|
|
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_lock(&nm_i->nid_list_lock);
|
2017-03-22 06:45:05 +00:00
|
|
|
|
|
|
|
if (build) {
|
|
|
|
/*
|
|
|
|
* Thread A Thread B
|
|
|
|
* - f2fs_create
|
|
|
|
* - f2fs_new_inode
|
|
|
|
* - alloc_nid
|
2017-09-29 05:59:35 +00:00
|
|
|
* - __insert_nid_to_list(PREALLOC_NID)
|
2017-03-22 06:45:05 +00:00
|
|
|
* - f2fs_balance_fs_bg
|
|
|
|
* - build_free_nids
|
|
|
|
* - __build_free_nids
|
|
|
|
* - scan_nat_page
|
|
|
|
* - add_free_nid
|
|
|
|
* - __lookup_nat_cache
|
|
|
|
* - f2fs_add_link
|
|
|
|
* - init_inode_metadata
|
|
|
|
* - new_inode_page
|
|
|
|
* - new_node_page
|
|
|
|
* - set_node_addr
|
|
|
|
* - alloc_nid_done
|
2017-09-29 05:59:35 +00:00
|
|
|
* - __remove_nid_from_list(PREALLOC_NID)
|
|
|
|
* - __insert_nid_to_list(FREE_NID)
|
2017-03-22 06:45:05 +00:00
|
|
|
*/
|
|
|
|
ne = __lookup_nat_cache(nm_i, nid);
|
|
|
|
if (ne && (!get_nat_flag(ne, IS_CHECKPOINTED) ||
|
|
|
|
nat_get_blkaddr(ne) != NULL_ADDR))
|
|
|
|
goto err_out;
|
|
|
|
|
|
|
|
e = __lookup_free_nid_list(nm_i, nid);
|
|
|
|
if (e) {
|
2017-09-29 05:59:35 +00:00
|
|
|
if (e->state == FREE_NID)
|
2017-03-22 06:45:05 +00:00
|
|
|
ret = true;
|
|
|
|
goto err_out;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
ret = true;
|
2017-10-28 11:03:37 +00:00
|
|
|
err = __insert_free_nid(sbi, i, FREE_NID);
|
2017-03-22 06:45:05 +00:00
|
|
|
err_out:
|
2017-11-22 08:07:23 +00:00
|
|
|
if (update) {
|
|
|
|
update_free_nid_bitmap(sbi, nid, ret, build);
|
|
|
|
if (!build)
|
|
|
|
nm_i->available_nids++;
|
|
|
|
}
|
2016-10-12 17:09:59 +00:00
|
|
|
spin_unlock(&nm_i->nid_list_lock);
|
|
|
|
radix_tree_preload_end();
|
2017-11-22 08:07:23 +00:00
|
|
|
|
2017-03-22 06:45:05 +00:00
|
|
|
if (err)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
kmem_cache_free(free_nid_slab, i);
|
2017-03-22 06:45:05 +00:00
|
|
|
return ret;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
static void remove_free_nid(struct f2fs_sb_info *sbi, nid_t nid)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
struct free_nid *i;
|
2014-04-02 00:55:00 +00:00
|
|
|
bool need_free = false;
|
|
|
|
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_lock(&nm_i->nid_list_lock);
|
2014-02-21 05:29:35 +00:00
|
|
|
i = __lookup_free_nid_list(nm_i, nid);
|
2017-09-29 05:59:35 +00:00
|
|
|
if (i && i->state == FREE_NID) {
|
2017-10-28 11:03:37 +00:00
|
|
|
__remove_free_nid(sbi, i, FREE_NID);
|
2014-04-02 00:55:00 +00:00
|
|
|
need_free = true;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_unlock(&nm_i->nid_list_lock);
|
2014-04-02 00:55:00 +00:00
|
|
|
|
|
|
|
if (need_free)
|
|
|
|
kmem_cache_free(free_nid_slab, i);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2014-04-16 01:47:06 +00:00
|
|
|
static void scan_nat_page(struct f2fs_sb_info *sbi,
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
struct page *nat_page, nid_t start_nid)
|
|
|
|
{
|
2014-04-16 01:47:06 +00:00
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
struct f2fs_nat_block *nat_blk = page_address(nat_page);
|
|
|
|
block_t blk_addr;
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
unsigned int nat_ofs = NAT_BLOCK_OFFSET(start_nid);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
int i;
|
|
|
|
|
2017-03-07 22:11:06 +00:00
|
|
|
__set_bit_le(nat_ofs, nm_i->nat_block_bitmap);
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
i = start_nid % NAT_ENTRY_PER_BLOCK;
|
|
|
|
|
|
|
|
for (; i < NAT_ENTRY_PER_BLOCK; i++, start_nid++) {
|
2013-12-05 09:15:22 +00:00
|
|
|
if (unlikely(start_nid >= nm_i->max_nid))
|
2013-03-15 23:34:37 +00:00
|
|
|
break;
|
2013-05-06 15:15:43 +00:00
|
|
|
|
|
|
|
blk_addr = le32_to_cpu(nat_blk->entries[i].block_addr);
|
2014-09-02 22:52:58 +00:00
|
|
|
f2fs_bug_on(sbi, blk_addr == NEW_ADDR);
|
2017-11-22 08:07:23 +00:00
|
|
|
if (blk_addr == NULL_ADDR) {
|
|
|
|
add_free_nid(sbi, start_nid, true, true);
|
|
|
|
} else {
|
|
|
|
spin_lock(&NM_I(sbi)->nid_list_lock);
|
|
|
|
update_free_nid_bitmap(sbi, start_nid, false, true);
|
|
|
|
spin_unlock(&NM_I(sbi)->nid_list_lock);
|
|
|
|
}
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2017-11-08 09:47:36 +00:00
|
|
|
static void scan_curseg_cache(struct f2fs_sb_info *sbi)
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
{
|
|
|
|
struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA);
|
|
|
|
struct f2fs_journal *journal = curseg->journal;
|
2017-11-08 09:47:36 +00:00
|
|
|
int i;
|
|
|
|
|
|
|
|
down_read(&curseg->journal_rwsem);
|
|
|
|
for (i = 0; i < nats_in_cursum(journal); i++) {
|
|
|
|
block_t addr;
|
|
|
|
nid_t nid;
|
|
|
|
|
|
|
|
addr = le32_to_cpu(nat_in_journal(journal, i).block_addr);
|
|
|
|
nid = le32_to_cpu(nid_in_journal(journal, i));
|
|
|
|
if (addr == NULL_ADDR)
|
2017-11-22 08:07:23 +00:00
|
|
|
add_free_nid(sbi, nid, true, false);
|
2017-11-08 09:47:36 +00:00
|
|
|
else
|
|
|
|
remove_free_nid(sbi, nid);
|
|
|
|
}
|
|
|
|
up_read(&curseg->journal_rwsem);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void scan_free_nid_bits(struct f2fs_sb_info *sbi)
|
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
unsigned int i, idx;
|
2017-11-07 11:14:24 +00:00
|
|
|
nid_t nid;
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
|
|
|
|
down_read(&nm_i->nat_tree_lock);
|
|
|
|
|
|
|
|
for (i = 0; i < nm_i->nat_blocks; i++) {
|
|
|
|
if (!test_bit_le(i, nm_i->nat_block_bitmap))
|
|
|
|
continue;
|
2017-03-01 09:09:07 +00:00
|
|
|
if (!nm_i->free_nid_count[i])
|
|
|
|
continue;
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
for (idx = 0; idx < NAT_ENTRY_PER_BLOCK; idx++) {
|
2017-11-07 11:14:24 +00:00
|
|
|
idx = find_next_bit_le(nm_i->free_nid_bitmap[i],
|
|
|
|
NAT_ENTRY_PER_BLOCK, idx);
|
|
|
|
if (idx >= NAT_ENTRY_PER_BLOCK)
|
|
|
|
break;
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
|
|
|
|
nid = i * NAT_ENTRY_PER_BLOCK + idx;
|
2017-11-22 08:07:23 +00:00
|
|
|
add_free_nid(sbi, nid, true, false);
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
|
2017-09-29 05:59:35 +00:00
|
|
|
if (nm_i->nid_cnt[FREE_NID] >= MAX_FREE_NIDS)
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
out:
|
2017-11-08 09:47:36 +00:00
|
|
|
scan_curseg_cache(sbi);
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
|
|
|
|
up_read(&nm_i->nat_tree_lock);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2017-02-09 18:38:09 +00:00
|
|
|
static void __build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
2013-05-06 15:15:42 +00:00
|
|
|
int i = 0;
|
2013-04-25 07:05:51 +00:00
|
|
|
nid_t nid = nm_i->next_scan_nid;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2017-04-26 07:56:52 +00:00
|
|
|
if (unlikely(nid >= nm_i->max_nid))
|
|
|
|
nid = 0;
|
|
|
|
|
2013-04-25 07:05:51 +00:00
|
|
|
/* Enough entries */
|
2017-09-29 05:59:35 +00:00
|
|
|
if (nm_i->nid_cnt[FREE_NID] >= NAT_ENTRY_PER_BLOCK)
|
2013-04-25 07:05:51 +00:00
|
|
|
return;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2016-10-11 14:31:35 +00:00
|
|
|
if (!sync && !available_free_memory(sbi, FREE_NIDS))
|
2013-04-25 07:05:51 +00:00
|
|
|
return;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
if (!mount) {
|
|
|
|
/* try to find free nids in free_nid_bitmap */
|
|
|
|
scan_free_nid_bits(sbi);
|
|
|
|
|
2017-11-07 03:04:33 +00:00
|
|
|
if (nm_i->nid_cnt[FREE_NID] >= NAT_ENTRY_PER_BLOCK)
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
return;
|
2017-02-09 18:38:09 +00:00
|
|
|
}
|
|
|
|
|
2013-04-25 07:05:51 +00:00
|
|
|
/* readahead nat pages to be scanned */
|
2015-10-12 09:05:59 +00:00
|
|
|
ra_meta_pages(sbi, NAT_BLOCK_OFFSET(nid), FREE_NID_PAGES,
|
|
|
|
META_NAT, true);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2016-08-04 18:38:25 +00:00
|
|
|
down_read(&nm_i->nat_tree_lock);
|
2016-01-02 17:19:41 +00:00
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
while (1) {
|
2017-11-17 08:13:38 +00:00
|
|
|
if (!test_bit_le(NAT_BLOCK_OFFSET(nid),
|
|
|
|
nm_i->nat_block_bitmap)) {
|
|
|
|
struct page *page = get_current_nat_page(sbi, nid);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2017-11-17 08:13:38 +00:00
|
|
|
scan_nat_page(sbi, page, nid);
|
|
|
|
f2fs_put_page(page, 1);
|
|
|
|
}
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
nid += (NAT_ENTRY_PER_BLOCK - (nid % NAT_ENTRY_PER_BLOCK));
|
2013-12-05 09:15:22 +00:00
|
|
|
if (unlikely(nid >= nm_i->max_nid))
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
nid = 0;
|
2013-04-25 07:05:51 +00:00
|
|
|
|
2015-07-24 10:26:26 +00:00
|
|
|
if (++i >= FREE_NID_PAGES)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2013-04-25 07:05:51 +00:00
|
|
|
/* go to the next free nat pages to find free nids abundantly */
|
|
|
|
nm_i->next_scan_nid = nid;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
/* find free nids from current sum_pages */
|
2017-11-08 09:47:36 +00:00
|
|
|
scan_curseg_cache(sbi);
|
2016-02-14 10:50:40 +00:00
|
|
|
|
2016-08-04 18:38:25 +00:00
|
|
|
up_read(&nm_i->nat_tree_lock);
|
2015-10-12 09:07:33 +00:00
|
|
|
|
|
|
|
ra_meta_pages(sbi, NAT_BLOCK_OFFSET(nm_i->next_scan_nid),
|
2015-10-12 09:08:48 +00:00
|
|
|
nm_i->ra_nid_pages, META_NAT, false);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2017-02-09 18:38:09 +00:00
|
|
|
void build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount)
|
2016-10-11 14:31:34 +00:00
|
|
|
{
|
|
|
|
mutex_lock(&NM_I(sbi)->build_lock);
|
2017-02-09 18:38:09 +00:00
|
|
|
__build_free_nids(sbi, sync, mount);
|
2016-10-11 14:31:34 +00:00
|
|
|
mutex_unlock(&NM_I(sbi)->build_lock);
|
|
|
|
}
|
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
/*
|
|
|
|
* If this function returns success, caller can obtain a new nid
|
|
|
|
* from second parameter of this function.
|
|
|
|
* The returned nid could be used ino as well as nid when inode is created.
|
|
|
|
*/
|
|
|
|
bool alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid)
|
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
struct free_nid *i = NULL;
|
|
|
|
retry:
|
2016-04-29 23:29:22 +00:00
|
|
|
#ifdef CONFIG_F2FS_FAULT_INJECTION
|
2017-02-25 03:08:28 +00:00
|
|
|
if (time_to_inject(sbi, FAULT_ALLOC_NID)) {
|
|
|
|
f2fs_show_injection_info(FAULT_ALLOC_NID);
|
2016-04-29 23:29:22 +00:00
|
|
|
return false;
|
2017-02-25 03:08:28 +00:00
|
|
|
}
|
2016-04-29 23:29:22 +00:00
|
|
|
#endif
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_lock(&nm_i->nid_list_lock);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2016-11-17 12:53:11 +00:00
|
|
|
if (unlikely(nm_i->available_nids == 0)) {
|
|
|
|
spin_unlock(&nm_i->nid_list_lock);
|
|
|
|
return false;
|
|
|
|
}
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2013-04-25 07:05:51 +00:00
|
|
|
/* We should not use stale free nids created by build_free_nids */
|
2017-09-29 05:59:35 +00:00
|
|
|
if (nm_i->nid_cnt[FREE_NID] && !on_build_free_nids(nm_i)) {
|
|
|
|
f2fs_bug_on(sbi, list_empty(&nm_i->free_nid_list));
|
|
|
|
i = list_first_entry(&nm_i->free_nid_list,
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
struct free_nid, list);
|
2013-04-25 07:05:51 +00:00
|
|
|
*nid = i->nid;
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
|
2017-10-28 11:03:37 +00:00
|
|
|
__move_free_nid(sbi, i, FREE_NID, PREALLOC_NID);
|
2016-11-17 12:53:11 +00:00
|
|
|
nm_i->available_nids--;
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
|
2017-03-13 12:10:41 +00:00
|
|
|
update_free_nid_bitmap(sbi, *nid, false, false);
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_unlock(&nm_i->nid_list_lock);
|
2013-04-25 07:05:51 +00:00
|
|
|
return true;
|
|
|
|
}
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_unlock(&nm_i->nid_list_lock);
|
2013-04-25 07:05:51 +00:00
|
|
|
|
|
|
|
/* Let's scan nat pages and its caches to get free nids */
|
2017-02-09 18:38:09 +00:00
|
|
|
build_free_nids(sbi, true, false);
|
2013-04-25 07:05:51 +00:00
|
|
|
goto retry;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2012-11-29 04:28:09 +00:00
|
|
|
/*
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
* alloc_nid() should be called prior to this function.
|
|
|
|
*/
|
|
|
|
void alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid)
|
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
struct free_nid *i;
|
|
|
|
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_lock(&nm_i->nid_list_lock);
|
2014-02-21 05:29:35 +00:00
|
|
|
i = __lookup_free_nid_list(nm_i, nid);
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
f2fs_bug_on(sbi, !i);
|
2017-10-28 11:03:37 +00:00
|
|
|
__remove_free_nid(sbi, i, PREALLOC_NID);
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_unlock(&nm_i->nid_list_lock);
|
2014-04-02 00:55:00 +00:00
|
|
|
|
|
|
|
kmem_cache_free(free_nid_slab, i);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2012-11-29 04:28:09 +00:00
|
|
|
/*
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
* alloc_nid() should be called prior to this function.
|
|
|
|
*/
|
|
|
|
void alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid)
|
|
|
|
{
|
2013-04-03 13:19:03 +00:00
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
struct free_nid *i;
|
2014-04-02 00:55:00 +00:00
|
|
|
bool need_free = false;
|
2013-04-03 13:19:03 +00:00
|
|
|
|
2013-08-14 12:57:27 +00:00
|
|
|
if (!nid)
|
|
|
|
return;
|
|
|
|
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_lock(&nm_i->nid_list_lock);
|
2014-02-21 05:29:35 +00:00
|
|
|
i = __lookup_free_nid_list(nm_i, nid);
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
f2fs_bug_on(sbi, !i);
|
|
|
|
|
2014-04-16 01:47:06 +00:00
|
|
|
if (!available_free_memory(sbi, FREE_NIDS)) {
|
2017-10-28 11:03:37 +00:00
|
|
|
__remove_free_nid(sbi, i, PREALLOC_NID);
|
2014-04-02 00:55:00 +00:00
|
|
|
need_free = true;
|
2013-05-06 15:15:41 +00:00
|
|
|
} else {
|
2017-10-28 11:03:37 +00:00
|
|
|
__move_free_nid(sbi, i, PREALLOC_NID, FREE_NID);
|
2013-05-06 15:15:41 +00:00
|
|
|
}
|
2016-11-17 12:53:11 +00:00
|
|
|
|
|
|
|
nm_i->available_nids++;
|
|
|
|
|
2017-03-13 12:10:41 +00:00
|
|
|
update_free_nid_bitmap(sbi, nid, true, false);
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_unlock(&nm_i->nid_list_lock);
|
2014-04-02 00:55:00 +00:00
|
|
|
|
|
|
|
if (need_free)
|
|
|
|
kmem_cache_free(free_nid_slab, i);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2015-07-28 10:33:46 +00:00
|
|
|
int try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink)
|
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
struct free_nid *i, *next;
|
|
|
|
int nr = nr_shrink;
|
|
|
|
|
2017-09-29 05:59:35 +00:00
|
|
|
if (nm_i->nid_cnt[FREE_NID] <= MAX_FREE_NIDS)
|
2016-06-16 23:41:49 +00:00
|
|
|
return 0;
|
|
|
|
|
2015-07-28 10:33:46 +00:00
|
|
|
if (!mutex_trylock(&nm_i->build_lock))
|
|
|
|
return 0;
|
|
|
|
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_lock(&nm_i->nid_list_lock);
|
2017-09-29 05:59:35 +00:00
|
|
|
list_for_each_entry_safe(i, next, &nm_i->free_nid_list, list) {
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
if (nr_shrink <= 0 ||
|
2017-09-29 05:59:35 +00:00
|
|
|
nm_i->nid_cnt[FREE_NID] <= MAX_FREE_NIDS)
|
2015-07-28 10:33:46 +00:00
|
|
|
break;
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
|
2017-10-28 11:03:37 +00:00
|
|
|
__remove_free_nid(sbi, i, FREE_NID);
|
2015-07-28 10:33:46 +00:00
|
|
|
kmem_cache_free(free_nid_slab, i);
|
|
|
|
nr_shrink--;
|
|
|
|
}
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_unlock(&nm_i->nid_list_lock);
|
2015-07-28 10:33:46 +00:00
|
|
|
mutex_unlock(&nm_i->build_lock);
|
|
|
|
|
|
|
|
return nr - nr_shrink;
|
|
|
|
}
|
|
|
|
|
2014-08-02 07:26:04 +00:00
|
|
|
void recover_inline_xattr(struct inode *inode, struct page *page)
|
2014-03-11 05:37:38 +00:00
|
|
|
{
|
|
|
|
void *src_addr, *dst_addr;
|
|
|
|
size_t inline_size;
|
|
|
|
struct page *ipage;
|
|
|
|
struct f2fs_inode *ri;
|
|
|
|
|
2014-09-02 22:31:18 +00:00
|
|
|
ipage = get_node_page(F2FS_I_SB(inode), inode->i_ino);
|
2014-09-02 22:52:58 +00:00
|
|
|
f2fs_bug_on(F2FS_I_SB(inode), IS_ERR(ipage));
|
2014-03-11 05:37:38 +00:00
|
|
|
|
2014-08-08 06:45:42 +00:00
|
|
|
ri = F2FS_INODE(page);
|
2018-01-03 10:03:04 +00:00
|
|
|
if (ri->i_inline & F2FS_INLINE_XATTR) {
|
|
|
|
set_inode_flag(inode, FI_INLINE_XATTR);
|
|
|
|
} else {
|
2016-05-20 17:13:22 +00:00
|
|
|
clear_inode_flag(inode, FI_INLINE_XATTR);
|
2014-08-08 06:45:42 +00:00
|
|
|
goto update_inode;
|
|
|
|
}
|
|
|
|
|
f2fs: support flexible inline xattr size
Now, in product, more and more features based on file encryption were
introduced, their demand of xattr space is increasing, however, inline
xattr has fixed-size of 200 bytes, once inline xattr space is full, new
increased xattr data would occupy additional xattr block which may bring
us more space usage and performance regression during persisting.
In order to resolve above issue, it's better to expand inline xattr size
flexibly according to user's requirement.
So this patch introduces new filesystem feature 'flexible inline xattr',
and new mount option 'inline_xattr_size=%u', once mkfs enables the
feature, we can use the option to make f2fs supporting flexible inline
xattr size.
To support this feature, we add extra attribute i_inline_xattr_size in
inode layout, indicating that how many space inline xattr borrows from
block address mapping space in inode layout, by this, we can easily
locate and store flexible-sized inline xattr data in inode.
Inode disk layout:
+----------------------+
| .i_mode |
| ... |
| .i_ext |
+----------------------+
| .i_extra_isize |
| .i_inline_xattr_size |-----------+
| ... | |
+----------------------+ |
| .i_addr | |
| - block address or | |
| - inline data | |
+----------------------+<---+ v
| inline xattr | +---inline xattr range
+----------------------+<---+
| .i_nid |
+----------------------+
| node_footer |
| (nid, ino, offset) |
+----------------------+
Note that, we have to cnosider backward compatibility which reserved
inline_data space, 200 bytes, all the time, reported by Sheng Yong.
Previous inline data or directory always reserved 200 bytes in inode layout,
even if inline_xattr is disabled. In order to keep inline_dentry's structure
for backward compatibility, we get the space back only from inline_data.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Reported-by: Sheng Yong <shengyong1@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-09-06 13:59:50 +00:00
|
|
|
dst_addr = inline_xattr_addr(inode, ipage);
|
|
|
|
src_addr = inline_xattr_addr(inode, page);
|
2014-03-11 05:37:38 +00:00
|
|
|
inline_size = inline_xattr_size(inode);
|
|
|
|
|
2016-01-20 15:43:51 +00:00
|
|
|
f2fs_wait_on_page_writeback(ipage, NODE, true);
|
2014-03-11 05:37:38 +00:00
|
|
|
memcpy(dst_addr, src_addr, inline_size);
|
2014-08-08 06:45:42 +00:00
|
|
|
update_inode:
|
2014-03-11 05:37:38 +00:00
|
|
|
update_inode(inode, ipage);
|
|
|
|
f2fs_put_page(ipage, 1);
|
|
|
|
}
|
|
|
|
|
2017-11-22 10:23:40 +00:00
|
|
|
int recover_xattr_data(struct inode *inode, struct page *page)
|
2014-01-28 03:25:06 +00:00
|
|
|
{
|
2014-09-02 22:31:18 +00:00
|
|
|
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
|
2014-01-28 03:25:06 +00:00
|
|
|
nid_t prev_xnid = F2FS_I(inode)->i_xattr_nid;
|
2017-07-18 01:48:12 +00:00
|
|
|
nid_t new_xnid;
|
|
|
|
struct dnode_of_data dn;
|
2014-01-28 03:25:06 +00:00
|
|
|
struct node_info ni;
|
f2fs: change recovery policy of xattr node block
Currently, if we call fsync after updating the xattr date belongs to the
file, f2fs needs to trigger checkpoint to keep xattr data consistent. But,
this policy cause low performance as checkpoint will block most foreground
operations and cause unneeded and unrelated IOs around checkpoint.
This patch will reuse regular file recovery policy for xattr node block,
so, we change to write xattr node block tagged with fsync flag to warm
area instead of cold area, and during recovery, we search warm node chain
for fsynced xattr block, and do the recovery.
So, for below application IO pattern, performance can be improved
obviously:
- touch file
- create/update/delete xattr entry in file
- fsync file
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-08 09:39:45 +00:00
|
|
|
struct page *xpage;
|
2014-01-28 03:25:06 +00:00
|
|
|
|
|
|
|
if (!prev_xnid)
|
|
|
|
goto recover_xnid;
|
|
|
|
|
f2fs: change recovery policy of xattr node block
Currently, if we call fsync after updating the xattr date belongs to the
file, f2fs needs to trigger checkpoint to keep xattr data consistent. But,
this policy cause low performance as checkpoint will block most foreground
operations and cause unneeded and unrelated IOs around checkpoint.
This patch will reuse regular file recovery policy for xattr node block,
so, we change to write xattr node block tagged with fsync flag to warm
area instead of cold area, and during recovery, we search warm node chain
for fsynced xattr block, and do the recovery.
So, for below application IO pattern, performance can be improved
obviously:
- touch file
- create/update/delete xattr entry in file
- fsync file
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-08 09:39:45 +00:00
|
|
|
/* 1: invalidate the previous xattr nid */
|
2014-01-28 03:25:06 +00:00
|
|
|
get_node_info(sbi, prev_xnid, &ni);
|
|
|
|
invalidate_blocks(sbi, ni.blk_addr);
|
f2fs: don't count inode block in in-memory inode.i_blocks
Previously, we count all inode consumed blocks including inode block,
xattr block, index block, data block into i_blocks, for other generic
filesystems, they won't count inode block into i_blocks, so for
userspace applications or quota system, they may detect incorrect block
count according to i_blocks value in inode.
This patch changes to count all blocks into inode.i_blocks excluding
inode block, for on-disk i_blocks, we keep counting inode block for
backward compatibility.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-05 17:11:31 +00:00
|
|
|
dec_valid_node_count(sbi, inode, false);
|
2014-03-20 12:52:53 +00:00
|
|
|
set_node_addr(sbi, &ni, NULL_ADDR, false);
|
2014-01-28 03:25:06 +00:00
|
|
|
|
|
|
|
recover_xnid:
|
f2fs: change recovery policy of xattr node block
Currently, if we call fsync after updating the xattr date belongs to the
file, f2fs needs to trigger checkpoint to keep xattr data consistent. But,
this policy cause low performance as checkpoint will block most foreground
operations and cause unneeded and unrelated IOs around checkpoint.
This patch will reuse regular file recovery policy for xattr node block,
so, we change to write xattr node block tagged with fsync flag to warm
area instead of cold area, and during recovery, we search warm node chain
for fsynced xattr block, and do the recovery.
So, for below application IO pattern, performance can be improved
obviously:
- touch file
- create/update/delete xattr entry in file
- fsync file
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-08 09:39:45 +00:00
|
|
|
/* 2: update xattr nid in inode */
|
2017-07-18 01:48:12 +00:00
|
|
|
if (!alloc_nid(sbi, &new_xnid))
|
|
|
|
return -ENOSPC;
|
|
|
|
|
|
|
|
set_new_dnode(&dn, inode, NULL, NULL, new_xnid);
|
|
|
|
xpage = new_node_page(&dn, XATTR_NODE_OFFSET);
|
|
|
|
if (IS_ERR(xpage)) {
|
|
|
|
alloc_nid_failed(sbi, new_xnid);
|
|
|
|
return PTR_ERR(xpage);
|
|
|
|
}
|
|
|
|
|
|
|
|
alloc_nid_done(sbi, new_xnid);
|
f2fs: change recovery policy of xattr node block
Currently, if we call fsync after updating the xattr date belongs to the
file, f2fs needs to trigger checkpoint to keep xattr data consistent. But,
this policy cause low performance as checkpoint will block most foreground
operations and cause unneeded and unrelated IOs around checkpoint.
This patch will reuse regular file recovery policy for xattr node block,
so, we change to write xattr node block tagged with fsync flag to warm
area instead of cold area, and during recovery, we search warm node chain
for fsynced xattr block, and do the recovery.
So, for below application IO pattern, performance can be improved
obviously:
- touch file
- create/update/delete xattr entry in file
- fsync file
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-08 09:39:45 +00:00
|
|
|
update_inode_page(inode);
|
|
|
|
|
|
|
|
/* 3: update and set xattr node page dirty */
|
2017-07-18 01:48:12 +00:00
|
|
|
memcpy(F2FS_NODE(xpage), F2FS_NODE(page), VALID_XATTR_BLOCK_SIZE);
|
2014-01-28 03:25:06 +00:00
|
|
|
|
f2fs: change recovery policy of xattr node block
Currently, if we call fsync after updating the xattr date belongs to the
file, f2fs needs to trigger checkpoint to keep xattr data consistent. But,
this policy cause low performance as checkpoint will block most foreground
operations and cause unneeded and unrelated IOs around checkpoint.
This patch will reuse regular file recovery policy for xattr node block,
so, we change to write xattr node block tagged with fsync flag to warm
area instead of cold area, and during recovery, we search warm node chain
for fsynced xattr block, and do the recovery.
So, for below application IO pattern, performance can be improved
obviously:
- touch file
- create/update/delete xattr entry in file
- fsync file
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-08 09:39:45 +00:00
|
|
|
set_page_dirty(xpage);
|
|
|
|
f2fs_put_page(xpage, 1);
|
2014-01-28 03:25:06 +00:00
|
|
|
|
f2fs: change recovery policy of xattr node block
Currently, if we call fsync after updating the xattr date belongs to the
file, f2fs needs to trigger checkpoint to keep xattr data consistent. But,
this policy cause low performance as checkpoint will block most foreground
operations and cause unneeded and unrelated IOs around checkpoint.
This patch will reuse regular file recovery policy for xattr node block,
so, we change to write xattr node block tagged with fsync flag to warm
area instead of cold area, and during recovery, we search warm node chain
for fsynced xattr block, and do the recovery.
So, for below application IO pattern, performance can be improved
obviously:
- touch file
- create/update/delete xattr entry in file
- fsync file
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-08 09:39:45 +00:00
|
|
|
return 0;
|
2014-01-28 03:25:06 +00:00
|
|
|
}
|
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
int recover_inode_page(struct f2fs_sb_info *sbi, struct page *page)
|
|
|
|
{
|
2013-12-26 07:30:41 +00:00
|
|
|
struct f2fs_inode *src, *dst;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
nid_t ino = ino_of_node(page);
|
|
|
|
struct node_info old_ni, new_ni;
|
|
|
|
struct page *ipage;
|
|
|
|
|
2014-04-18 06:21:04 +00:00
|
|
|
get_node_info(sbi, ino, &old_ni);
|
|
|
|
|
|
|
|
if (unlikely(old_ni.blk_addr != NULL_ADDR))
|
|
|
|
return -EINVAL;
|
2016-09-09 23:59:39 +00:00
|
|
|
retry:
|
2016-04-29 23:11:53 +00:00
|
|
|
ipage = f2fs_grab_cache_page(NODE_MAPPING(sbi), ino, false);
|
2016-09-09 23:59:39 +00:00
|
|
|
if (!ipage) {
|
|
|
|
congestion_wait(BLK_RW_ASYNC, HZ/50);
|
|
|
|
goto retry;
|
|
|
|
}
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2014-08-06 14:22:50 +00:00
|
|
|
/* Should not use this inode from free nid list */
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
remove_free_nid(sbi, ino);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2016-07-01 01:49:15 +00:00
|
|
|
if (!PageUptodate(ipage))
|
|
|
|
SetPageUptodate(ipage);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
fill_node_footer(ipage, ino, ino, 0, true);
|
2018-03-09 06:24:22 +00:00
|
|
|
set_cold_node(page, false);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2013-12-26 07:30:41 +00:00
|
|
|
src = F2FS_INODE(page);
|
|
|
|
dst = F2FS_INODE(ipage);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2013-12-26 07:30:41 +00:00
|
|
|
memcpy(dst, src, (unsigned long)&src->i_ext - (unsigned long)src);
|
|
|
|
dst->i_size = 0;
|
|
|
|
dst->i_blocks = cpu_to_le64(1);
|
|
|
|
dst->i_links = cpu_to_le32(1);
|
|
|
|
dst->i_xattr_nid = 0;
|
2017-07-18 16:19:06 +00:00
|
|
|
dst->i_inline = src->i_inline & (F2FS_INLINE_XATTR | F2FS_EXTRA_ATTR);
|
2017-07-25 16:01:41 +00:00
|
|
|
if (dst->i_inline & F2FS_EXTRA_ATTR) {
|
2017-07-18 16:19:06 +00:00
|
|
|
dst->i_extra_isize = src->i_extra_isize;
|
f2fs: support flexible inline xattr size
Now, in product, more and more features based on file encryption were
introduced, their demand of xattr space is increasing, however, inline
xattr has fixed-size of 200 bytes, once inline xattr space is full, new
increased xattr data would occupy additional xattr block which may bring
us more space usage and performance regression during persisting.
In order to resolve above issue, it's better to expand inline xattr size
flexibly according to user's requirement.
So this patch introduces new filesystem feature 'flexible inline xattr',
and new mount option 'inline_xattr_size=%u', once mkfs enables the
feature, we can use the option to make f2fs supporting flexible inline
xattr size.
To support this feature, we add extra attribute i_inline_xattr_size in
inode layout, indicating that how many space inline xattr borrows from
block address mapping space in inode layout, by this, we can easily
locate and store flexible-sized inline xattr data in inode.
Inode disk layout:
+----------------------+
| .i_mode |
| ... |
| .i_ext |
+----------------------+
| .i_extra_isize |
| .i_inline_xattr_size |-----------+
| ... | |
+----------------------+ |
| .i_addr | |
| - block address or | |
| - inline data | |
+----------------------+<---+ v
| inline xattr | +---inline xattr range
+----------------------+<---+
| .i_nid |
+----------------------+
| node_footer |
| (nid, ino, offset) |
+----------------------+
Note that, we have to cnosider backward compatibility which reserved
inline_data space, 200 bytes, all the time, reported by Sheng Yong.
Previous inline data or directory always reserved 200 bytes in inode layout,
even if inline_xattr is disabled. In order to keep inline_dentry's structure
for backward compatibility, we get the space back only from inline_data.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Reported-by: Sheng Yong <shengyong1@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-09-06 13:59:50 +00:00
|
|
|
|
|
|
|
if (f2fs_sb_has_flexible_inline_xattr(sbi->sb) &&
|
|
|
|
F2FS_FITS_IN_INODE(src, le16_to_cpu(src->i_extra_isize),
|
|
|
|
i_inline_xattr_size))
|
|
|
|
dst->i_inline_xattr_size = src->i_inline_xattr_size;
|
|
|
|
|
2017-07-25 16:01:41 +00:00
|
|
|
if (f2fs_sb_has_project_quota(sbi->sb) &&
|
|
|
|
F2FS_FITS_IN_INODE(src, le16_to_cpu(src->i_extra_isize),
|
|
|
|
i_projid))
|
|
|
|
dst->i_projid = src->i_projid;
|
|
|
|
}
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
new_ni = old_ni;
|
|
|
|
new_ni.ino = ino;
|
|
|
|
|
2017-07-08 16:13:07 +00:00
|
|
|
if (unlikely(inc_valid_node_count(sbi, NULL, true)))
|
2013-05-14 06:47:43 +00:00
|
|
|
WARN_ON(1);
|
2014-03-20 12:52:53 +00:00
|
|
|
set_node_addr(sbi, &new_ni, NEW_ADDR, false);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
inc_valid_inode_count(sbi);
|
2014-08-08 00:04:24 +00:00
|
|
|
set_page_dirty(ipage);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
f2fs_put_page(ipage, 1);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2017-12-06 03:31:29 +00:00
|
|
|
void restore_node_summary(struct f2fs_sb_info *sbi,
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
unsigned int segno, struct f2fs_summary_block *sum)
|
|
|
|
{
|
|
|
|
struct f2fs_node *rn;
|
|
|
|
struct f2fs_summary *sum_entry;
|
|
|
|
block_t addr;
|
2014-12-18 10:29:05 +00:00
|
|
|
int i, idx, last_offset, nrpages;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
/* scan the node segment */
|
|
|
|
last_offset = sbi->blocks_per_seg;
|
|
|
|
addr = START_BLOCK(sbi, segno);
|
|
|
|
sum_entry = &sum->entries[0];
|
|
|
|
|
2014-12-18 10:29:05 +00:00
|
|
|
for (i = 0; i < last_offset; i += nrpages, addr += nrpages) {
|
2016-10-18 18:07:45 +00:00
|
|
|
nrpages = min(last_offset - i, BIO_MAX_PAGES);
|
2013-03-08 12:29:23 +00:00
|
|
|
|
2014-08-06 14:22:50 +00:00
|
|
|
/* readahead node pages */
|
2015-10-12 09:05:59 +00:00
|
|
|
ra_meta_pages(sbi, addr, nrpages, META_POR, true);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2014-12-18 10:29:05 +00:00
|
|
|
for (idx = addr; idx < addr + nrpages; idx++) {
|
2015-10-12 09:04:21 +00:00
|
|
|
struct page *page = get_tmp_page(sbi, idx);
|
2013-11-22 07:48:54 +00:00
|
|
|
|
2014-12-18 10:29:05 +00:00
|
|
|
rn = F2FS_NODE(page);
|
|
|
|
sum_entry->nid = rn->footer.nid;
|
|
|
|
sum_entry->version = 0;
|
|
|
|
sum_entry->ofs_in_node = 0;
|
|
|
|
sum_entry++;
|
|
|
|
f2fs_put_page(page, 1);
|
2013-11-22 07:48:54 +00:00
|
|
|
}
|
2014-05-27 00:41:07 +00:00
|
|
|
|
2014-12-18 10:29:05 +00:00
|
|
|
invalidate_mapping_pages(META_MAPPING(sbi), addr,
|
2014-05-27 00:41:07 +00:00
|
|
|
addr + nrpages);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
static void remove_nats_in_journal(struct f2fs_sb_info *sbi)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA);
|
f2fs: split journal cache from curseg cache
In curseg cache, f2fs caches two different parts:
- datas of current summay block, i.e. summary entries, footer info.
- journal info, i.e. sparse nat/sit entries or io stat info.
With this approach, 1) it may cause higher lock contention when we access
or update both of the parts of cache since we use the same mutex lock
curseg_mutex to protect the cache. 2) current summary block with last
journal info will be writebacked into device as a normal summary block
when flushing, however, we treat journal info as valid one only in current
summary, so most normal summary blocks contain junk journal data, it wastes
remaining space of summary block.
So, in order to fix above issues, we split curseg cache into two parts:
a) current summary block, protected by original mutex lock curseg_mutex
b) journal cache, protected by newly introduced r/w semaphore journal_rwsem
When loading curseg cache during ->mount, we store summary info and
journal info into different caches; When doing checkpoint, we combine
datas of two cache into current summary block for persisting.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-19 10:08:46 +00:00
|
|
|
struct f2fs_journal *journal = curseg->journal;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
int i;
|
|
|
|
|
f2fs: split journal cache from curseg cache
In curseg cache, f2fs caches two different parts:
- datas of current summay block, i.e. summary entries, footer info.
- journal info, i.e. sparse nat/sit entries or io stat info.
With this approach, 1) it may cause higher lock contention when we access
or update both of the parts of cache since we use the same mutex lock
curseg_mutex to protect the cache. 2) current summary block with last
journal info will be writebacked into device as a normal summary block
when flushing, however, we treat journal info as valid one only in current
summary, so most normal summary blocks contain junk journal data, it wastes
remaining space of summary block.
So, in order to fix above issues, we split curseg cache into two parts:
a) current summary block, protected by original mutex lock curseg_mutex
b) journal cache, protected by newly introduced r/w semaphore journal_rwsem
When loading curseg cache during ->mount, we store summary info and
journal info into different caches; When doing checkpoint, we combine
datas of two cache into current summary block for persisting.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-19 10:08:46 +00:00
|
|
|
down_write(&curseg->journal_rwsem);
|
2016-02-14 10:50:40 +00:00
|
|
|
for (i = 0; i < nats_in_cursum(journal); i++) {
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
struct nat_entry *ne;
|
|
|
|
struct f2fs_nat_entry raw_ne;
|
2016-02-14 10:50:40 +00:00
|
|
|
nid_t nid = le32_to_cpu(nid_in_journal(journal, i));
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2016-02-14 10:50:40 +00:00
|
|
|
raw_ne = nat_in_journal(journal, i);
|
2014-12-05 18:39:49 +00:00
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
ne = __lookup_nat_cache(nm_i, nid);
|
|
|
|
if (!ne) {
|
2017-11-10 21:36:51 +00:00
|
|
|
ne = __alloc_nat_entry(nid, true);
|
|
|
|
__init_nat_entry(nm_i, ne, &raw_ne, true);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
2016-11-17 12:53:11 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* if a free nat in journal has not been used after last
|
|
|
|
* checkpoint, we should remove it from available nids,
|
|
|
|
* since later we will add it again.
|
|
|
|
*/
|
|
|
|
if (!get_nat_flag(ne, IS_DIRTY) &&
|
|
|
|
le32_to_cpu(raw_ne.block_addr) == NULL_ADDR) {
|
|
|
|
spin_lock(&nm_i->nid_list_lock);
|
|
|
|
nm_i->available_nids--;
|
|
|
|
spin_unlock(&nm_i->nid_list_lock);
|
|
|
|
}
|
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
__set_nat_cache_dirty(nm_i, ne);
|
|
|
|
}
|
2016-02-14 10:50:40 +00:00
|
|
|
update_nats_in_cursum(journal, -i);
|
f2fs: split journal cache from curseg cache
In curseg cache, f2fs caches two different parts:
- datas of current summay block, i.e. summary entries, footer info.
- journal info, i.e. sparse nat/sit entries or io stat info.
With this approach, 1) it may cause higher lock contention when we access
or update both of the parts of cache since we use the same mutex lock
curseg_mutex to protect the cache. 2) current summary block with last
journal info will be writebacked into device as a normal summary block
when flushing, however, we treat journal info as valid one only in current
summary, so most normal summary blocks contain junk journal data, it wastes
remaining space of summary block.
So, in order to fix above issues, we split curseg cache into two parts:
a) current summary block, protected by original mutex lock curseg_mutex
b) journal cache, protected by newly introduced r/w semaphore journal_rwsem
When loading curseg cache during ->mount, we store summary info and
journal info into different caches; When doing checkpoint, we combine
datas of two cache into current summary block for persisting.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-19 10:08:46 +00:00
|
|
|
up_write(&curseg->journal_rwsem);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2014-09-22 18:40:48 +00:00
|
|
|
static void __adjust_nat_entry_set(struct nat_entry_set *nes,
|
|
|
|
struct list_head *head, int max)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
2014-09-22 18:40:48 +00:00
|
|
|
struct nat_entry_set *cur;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2014-09-22 18:40:48 +00:00
|
|
|
if (nes->entry_cnt >= max)
|
|
|
|
goto add_out;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2014-09-22 18:40:48 +00:00
|
|
|
list_for_each_entry(cur, head, set_list) {
|
|
|
|
if (cur->entry_cnt >= nes->entry_cnt) {
|
|
|
|
list_add(&nes->set_list, cur->set_list.prev);
|
|
|
|
return;
|
|
|
|
}
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
}
|
2014-09-22 18:40:48 +00:00
|
|
|
add_out:
|
|
|
|
list_add_tail(&nes->set_list, head);
|
|
|
|
}
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2017-03-10 17:39:57 +00:00
|
|
|
static void __update_nat_bits(struct f2fs_sb_info *sbi, nid_t start_nid,
|
2017-02-09 18:38:09 +00:00
|
|
|
struct page *page)
|
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
unsigned int nat_index = start_nid / NAT_ENTRY_PER_BLOCK;
|
|
|
|
struct f2fs_nat_block *nat_blk = page_address(page);
|
|
|
|
int valid = 0;
|
2017-10-30 07:19:48 +00:00
|
|
|
int i = 0;
|
2017-02-09 18:38:09 +00:00
|
|
|
|
|
|
|
if (!enabled_nat_bits(sbi, NULL))
|
|
|
|
return;
|
|
|
|
|
2017-10-30 07:19:48 +00:00
|
|
|
if (nat_index == 0) {
|
|
|
|
valid = 1;
|
|
|
|
i = 1;
|
|
|
|
}
|
|
|
|
for (; i < NAT_ENTRY_PER_BLOCK; i++) {
|
|
|
|
if (nat_blk->entries[i].block_addr != NULL_ADDR)
|
2017-02-09 18:38:09 +00:00
|
|
|
valid++;
|
|
|
|
}
|
|
|
|
if (valid == 0) {
|
2017-03-07 22:11:06 +00:00
|
|
|
__set_bit_le(nat_index, nm_i->empty_nat_bits);
|
|
|
|
__clear_bit_le(nat_index, nm_i->full_nat_bits);
|
2017-02-09 18:38:09 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2017-03-07 22:11:06 +00:00
|
|
|
__clear_bit_le(nat_index, nm_i->empty_nat_bits);
|
2017-02-09 18:38:09 +00:00
|
|
|
if (valid == NAT_ENTRY_PER_BLOCK)
|
2017-03-07 22:11:06 +00:00
|
|
|
__set_bit_le(nat_index, nm_i->full_nat_bits);
|
2017-02-09 18:38:09 +00:00
|
|
|
else
|
2017-03-07 22:11:06 +00:00
|
|
|
__clear_bit_le(nat_index, nm_i->full_nat_bits);
|
2017-02-09 18:38:09 +00:00
|
|
|
}
|
|
|
|
|
2014-09-22 18:40:48 +00:00
|
|
|
static void __flush_nat_entry_set(struct f2fs_sb_info *sbi,
|
2017-02-09 18:38:09 +00:00
|
|
|
struct nat_entry_set *set, struct cp_control *cpc)
|
2014-09-22 18:40:48 +00:00
|
|
|
{
|
|
|
|
struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA);
|
f2fs: split journal cache from curseg cache
In curseg cache, f2fs caches two different parts:
- datas of current summay block, i.e. summary entries, footer info.
- journal info, i.e. sparse nat/sit entries or io stat info.
With this approach, 1) it may cause higher lock contention when we access
or update both of the parts of cache since we use the same mutex lock
curseg_mutex to protect the cache. 2) current summary block with last
journal info will be writebacked into device as a normal summary block
when flushing, however, we treat journal info as valid one only in current
summary, so most normal summary blocks contain junk journal data, it wastes
remaining space of summary block.
So, in order to fix above issues, we split curseg cache into two parts:
a) current summary block, protected by original mutex lock curseg_mutex
b) journal cache, protected by newly introduced r/w semaphore journal_rwsem
When loading curseg cache during ->mount, we store summary info and
journal info into different caches; When doing checkpoint, we combine
datas of two cache into current summary block for persisting.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-19 10:08:46 +00:00
|
|
|
struct f2fs_journal *journal = curseg->journal;
|
2014-09-22 18:40:48 +00:00
|
|
|
nid_t start_nid = set->set * NAT_ENTRY_PER_BLOCK;
|
|
|
|
bool to_journal = true;
|
|
|
|
struct f2fs_nat_block *nat_blk;
|
|
|
|
struct nat_entry *ne, *cur;
|
|
|
|
struct page *page = NULL;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
/*
|
|
|
|
* there are two steps to flush nat entries:
|
|
|
|
* #1, flush nat entries to journal in current hot data summary block.
|
|
|
|
* #2, flush nat entries to nat page.
|
|
|
|
*/
|
2017-02-09 18:38:09 +00:00
|
|
|
if (enabled_nat_bits(sbi, cpc) ||
|
|
|
|
!__has_cursum_space(journal, set->entry_cnt, NAT_JOURNAL))
|
2014-09-22 18:40:48 +00:00
|
|
|
to_journal = false;
|
|
|
|
|
|
|
|
if (to_journal) {
|
f2fs: split journal cache from curseg cache
In curseg cache, f2fs caches two different parts:
- datas of current summay block, i.e. summary entries, footer info.
- journal info, i.e. sparse nat/sit entries or io stat info.
With this approach, 1) it may cause higher lock contention when we access
or update both of the parts of cache since we use the same mutex lock
curseg_mutex to protect the cache. 2) current summary block with last
journal info will be writebacked into device as a normal summary block
when flushing, however, we treat journal info as valid one only in current
summary, so most normal summary blocks contain junk journal data, it wastes
remaining space of summary block.
So, in order to fix above issues, we split curseg cache into two parts:
a) current summary block, protected by original mutex lock curseg_mutex
b) journal cache, protected by newly introduced r/w semaphore journal_rwsem
When loading curseg cache during ->mount, we store summary info and
journal info into different caches; When doing checkpoint, we combine
datas of two cache into current summary block for persisting.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-19 10:08:46 +00:00
|
|
|
down_write(&curseg->journal_rwsem);
|
2014-09-22 18:40:48 +00:00
|
|
|
} else {
|
|
|
|
page = get_next_nat_page(sbi, start_nid);
|
|
|
|
nat_blk = page_address(page);
|
|
|
|
f2fs_bug_on(sbi, !nat_blk);
|
|
|
|
}
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
|
2014-09-22 18:40:48 +00:00
|
|
|
/* flush dirty nats in nat entry set */
|
|
|
|
list_for_each_entry_safe(ne, cur, &set->entry_list, list) {
|
|
|
|
struct f2fs_nat_entry *raw_ne;
|
|
|
|
nid_t nid = nat_get_nid(ne);
|
|
|
|
int offset;
|
|
|
|
|
2017-06-05 10:29:08 +00:00
|
|
|
f2fs_bug_on(sbi, nat_get_blkaddr(ne) == NEW_ADDR);
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
|
|
|
|
if (to_journal) {
|
2016-02-14 10:50:40 +00:00
|
|
|
offset = lookup_journal_in_cursum(journal,
|
2014-09-22 18:40:48 +00:00
|
|
|
NAT_JOURNAL, nid, 1);
|
|
|
|
f2fs_bug_on(sbi, offset < 0);
|
2016-02-14 10:50:40 +00:00
|
|
|
raw_ne = &nat_in_journal(journal, offset);
|
|
|
|
nid_in_journal(journal, offset) = cpu_to_le32(nid);
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
} else {
|
2014-09-22 18:40:48 +00:00
|
|
|
raw_ne = &nat_blk->entries[nid - start_nid];
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
2014-09-22 18:40:48 +00:00
|
|
|
raw_nat_from_node_info(raw_ne, &ne->ni);
|
|
|
|
nat_reset_flag(ne);
|
2017-02-28 13:34:47 +00:00
|
|
|
__clear_nat_cache_dirty(NM_I(sbi), set, ne);
|
2016-11-17 12:53:11 +00:00
|
|
|
if (nat_get_blkaddr(ne) == NULL_ADDR) {
|
2017-11-22 08:07:23 +00:00
|
|
|
add_free_nid(sbi, nid, false, true);
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
} else {
|
|
|
|
spin_lock(&NM_I(sbi)->nid_list_lock);
|
2017-03-13 12:10:41 +00:00
|
|
|
update_free_nid_bitmap(sbi, nid, false, false);
|
2016-11-17 12:53:11 +00:00
|
|
|
spin_unlock(&NM_I(sbi)->nid_list_lock);
|
|
|
|
}
|
2014-09-22 18:40:48 +00:00
|
|
|
}
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2017-02-09 18:38:09 +00:00
|
|
|
if (to_journal) {
|
f2fs: split journal cache from curseg cache
In curseg cache, f2fs caches two different parts:
- datas of current summay block, i.e. summary entries, footer info.
- journal info, i.e. sparse nat/sit entries or io stat info.
With this approach, 1) it may cause higher lock contention when we access
or update both of the parts of cache since we use the same mutex lock
curseg_mutex to protect the cache. 2) current summary block with last
journal info will be writebacked into device as a normal summary block
when flushing, however, we treat journal info as valid one only in current
summary, so most normal summary blocks contain junk journal data, it wastes
remaining space of summary block.
So, in order to fix above issues, we split curseg cache into two parts:
a) current summary block, protected by original mutex lock curseg_mutex
b) journal cache, protected by newly introduced r/w semaphore journal_rwsem
When loading curseg cache during ->mount, we store summary info and
journal info into different caches; When doing checkpoint, we combine
datas of two cache into current summary block for persisting.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-19 10:08:46 +00:00
|
|
|
up_write(&curseg->journal_rwsem);
|
2017-02-09 18:38:09 +00:00
|
|
|
} else {
|
|
|
|
__update_nat_bits(sbi, start_nid, page);
|
2014-09-22 18:40:48 +00:00
|
|
|
f2fs_put_page(page, 1);
|
2017-02-09 18:38:09 +00:00
|
|
|
}
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
|
2017-03-13 12:22:18 +00:00
|
|
|
/* Allow dirty nats by node block allocation in write_begin */
|
|
|
|
if (!set->entry_cnt) {
|
|
|
|
radix_tree_delete(&NM_I(sbi)->nat_set_root, set->set);
|
|
|
|
kmem_cache_free(nat_entry_set_slab, set);
|
|
|
|
}
|
2014-09-22 18:40:48 +00:00
|
|
|
}
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
|
2014-09-22 18:40:48 +00:00
|
|
|
/*
|
|
|
|
* This function is called during the checkpointing process.
|
|
|
|
*/
|
2017-02-09 18:38:09 +00:00
|
|
|
void flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc)
|
2014-09-22 18:40:48 +00:00
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA);
|
f2fs: split journal cache from curseg cache
In curseg cache, f2fs caches two different parts:
- datas of current summay block, i.e. summary entries, footer info.
- journal info, i.e. sparse nat/sit entries or io stat info.
With this approach, 1) it may cause higher lock contention when we access
or update both of the parts of cache since we use the same mutex lock
curseg_mutex to protect the cache. 2) current summary block with last
journal info will be writebacked into device as a normal summary block
when flushing, however, we treat journal info as valid one only in current
summary, so most normal summary blocks contain junk journal data, it wastes
remaining space of summary block.
So, in order to fix above issues, we split curseg cache into two parts:
a) current summary block, protected by original mutex lock curseg_mutex
b) journal cache, protected by newly introduced r/w semaphore journal_rwsem
When loading curseg cache during ->mount, we store summary info and
journal info into different caches; When doing checkpoint, we combine
datas of two cache into current summary block for persisting.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-19 10:08:46 +00:00
|
|
|
struct f2fs_journal *journal = curseg->journal;
|
2015-01-07 18:47:57 +00:00
|
|
|
struct nat_entry_set *setvec[SETVEC_SIZE];
|
2014-09-22 18:40:48 +00:00
|
|
|
struct nat_entry_set *set, *tmp;
|
|
|
|
unsigned int found;
|
|
|
|
nid_t set_idx = 0;
|
|
|
|
LIST_HEAD(sets);
|
|
|
|
|
2014-11-25 03:44:23 +00:00
|
|
|
if (!nm_i->dirty_nat_cnt)
|
|
|
|
return;
|
2016-01-02 17:19:41 +00:00
|
|
|
|
2016-08-04 18:38:25 +00:00
|
|
|
down_write(&nm_i->nat_tree_lock);
|
2016-01-02 17:19:41 +00:00
|
|
|
|
2014-09-22 18:40:48 +00:00
|
|
|
/*
|
|
|
|
* if there are no enough space in journal to store dirty nat
|
|
|
|
* entries, remove all entries from journal and merge them
|
|
|
|
* into nat entry set.
|
|
|
|
*/
|
2017-02-28 05:28:53 +00:00
|
|
|
if (enabled_nat_bits(sbi, cpc) ||
|
2017-02-09 18:38:09 +00:00
|
|
|
!__has_cursum_space(journal, nm_i->dirty_nat_cnt, NAT_JOURNAL))
|
2014-09-22 18:40:48 +00:00
|
|
|
remove_nats_in_journal(sbi);
|
|
|
|
|
|
|
|
while ((found = __gang_lookup_nat_set(nm_i,
|
2015-01-07 18:47:57 +00:00
|
|
|
set_idx, SETVEC_SIZE, setvec))) {
|
2014-09-22 18:40:48 +00:00
|
|
|
unsigned idx;
|
|
|
|
set_idx = setvec[found - 1]->set + 1;
|
|
|
|
for (idx = 0; idx < found; idx++)
|
|
|
|
__adjust_nat_entry_set(setvec[idx], &sets,
|
2016-02-14 10:50:40 +00:00
|
|
|
MAX_NAT_JENTRIES(journal));
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
|
2014-09-22 18:40:48 +00:00
|
|
|
/* flush dirty nats in nat entry set */
|
|
|
|
list_for_each_entry_safe(set, tmp, &sets, set_list)
|
2017-02-09 18:38:09 +00:00
|
|
|
__flush_nat_entry_set(sbi, set, cpc);
|
2014-09-22 18:40:48 +00:00
|
|
|
|
2016-08-04 18:38:25 +00:00
|
|
|
up_write(&nm_i->nat_tree_lock);
|
2017-03-13 12:22:18 +00:00
|
|
|
/* Allow dirty nats by node block allocation in write_begin */
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
2017-02-09 18:38:09 +00:00
|
|
|
static int __get_nat_bitmaps(struct f2fs_sb_info *sbi)
|
|
|
|
{
|
|
|
|
struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
unsigned int nat_bits_bytes = nm_i->nat_blocks / BITS_PER_BYTE;
|
|
|
|
unsigned int i;
|
|
|
|
__u64 cp_ver = cur_cp_version(ckpt);
|
|
|
|
block_t nat_bits_addr;
|
|
|
|
|
|
|
|
if (!enabled_nat_bits(sbi, NULL))
|
|
|
|
return 0;
|
|
|
|
|
2018-03-20 15:08:29 +00:00
|
|
|
nm_i->nat_bits_blocks = F2FS_BLK_ALIGN((nat_bits_bytes << 1) + 8);
|
2017-11-30 11:28:17 +00:00
|
|
|
nm_i->nat_bits = f2fs_kzalloc(sbi,
|
|
|
|
nm_i->nat_bits_blocks << F2FS_BLKSIZE_BITS, GFP_KERNEL);
|
2017-02-09 18:38:09 +00:00
|
|
|
if (!nm_i->nat_bits)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
nat_bits_addr = __start_cp_addr(sbi) + sbi->blocks_per_seg -
|
|
|
|
nm_i->nat_bits_blocks;
|
|
|
|
for (i = 0; i < nm_i->nat_bits_blocks; i++) {
|
|
|
|
struct page *page = get_meta_page(sbi, nat_bits_addr++);
|
|
|
|
|
|
|
|
memcpy(nm_i->nat_bits + (i << F2FS_BLKSIZE_BITS),
|
|
|
|
page_address(page), F2FS_BLKSIZE);
|
|
|
|
f2fs_put_page(page, 1);
|
|
|
|
}
|
|
|
|
|
2017-02-25 11:53:39 +00:00
|
|
|
cp_ver |= (cur_cp_crc(ckpt) << 32);
|
2017-02-09 18:38:09 +00:00
|
|
|
if (cpu_to_le64(cp_ver) != *(__le64 *)nm_i->nat_bits) {
|
|
|
|
disable_nat_bits(sbi, true);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
nm_i->full_nat_bits = nm_i->nat_bits + 8;
|
|
|
|
nm_i->empty_nat_bits = nm_i->full_nat_bits + nat_bits_bytes;
|
|
|
|
|
|
|
|
f2fs_msg(sbi->sb, KERN_NOTICE, "Found nat_bits in checkpoint");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2017-05-17 02:48:48 +00:00
|
|
|
static inline void load_free_nid_bitmap(struct f2fs_sb_info *sbi)
|
2017-03-08 12:07:49 +00:00
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
unsigned int i = 0;
|
|
|
|
nid_t nid, last_nid;
|
|
|
|
|
|
|
|
if (!enabled_nat_bits(sbi, NULL))
|
|
|
|
return;
|
|
|
|
|
|
|
|
for (i = 0; i < nm_i->nat_blocks; i++) {
|
|
|
|
i = find_next_bit_le(nm_i->empty_nat_bits, nm_i->nat_blocks, i);
|
|
|
|
if (i >= nm_i->nat_blocks)
|
|
|
|
break;
|
|
|
|
|
|
|
|
__set_bit_le(i, nm_i->nat_block_bitmap);
|
|
|
|
|
|
|
|
nid = i * NAT_ENTRY_PER_BLOCK;
|
2017-11-02 03:02:52 +00:00
|
|
|
last_nid = nid + NAT_ENTRY_PER_BLOCK;
|
2017-03-08 12:07:49 +00:00
|
|
|
|
2017-03-13 12:10:41 +00:00
|
|
|
spin_lock(&NM_I(sbi)->nid_list_lock);
|
2017-03-08 12:07:49 +00:00
|
|
|
for (; nid < last_nid; nid++)
|
2017-03-13 12:10:41 +00:00
|
|
|
update_free_nid_bitmap(sbi, nid, true, true);
|
|
|
|
spin_unlock(&NM_I(sbi)->nid_list_lock);
|
2017-03-08 12:07:49 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
for (i = 0; i < nm_i->nat_blocks; i++) {
|
|
|
|
i = find_next_bit_le(nm_i->full_nat_bits, nm_i->nat_blocks, i);
|
|
|
|
if (i >= nm_i->nat_blocks)
|
|
|
|
break;
|
|
|
|
|
|
|
|
__set_bit_le(i, nm_i->nat_block_bitmap);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
static int init_node_manager(struct f2fs_sb_info *sbi)
|
|
|
|
{
|
|
|
|
struct f2fs_super_block *sb_raw = F2FS_RAW_SUPER(sbi);
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
unsigned char *version_bitmap;
|
2017-02-09 18:38:09 +00:00
|
|
|
unsigned int nat_segs;
|
|
|
|
int err;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
nm_i->nat_blkaddr = le32_to_cpu(sb_raw->nat_blkaddr);
|
|
|
|
|
|
|
|
/* segment_count_nat includes pair segment so divide to 2. */
|
|
|
|
nat_segs = le32_to_cpu(sb_raw->segment_count_nat) >> 1;
|
2017-02-09 18:38:09 +00:00
|
|
|
nm_i->nat_blocks = nat_segs << le32_to_cpu(sb_raw->log_blocks_per_seg);
|
|
|
|
nm_i->max_nid = NAT_ENTRY_PER_BLOCK * nm_i->nat_blocks;
|
2014-04-18 02:14:37 +00:00
|
|
|
|
2014-02-17 03:44:20 +00:00
|
|
|
/* not used nids: 0, node, meta, (and root counted as valid node) */
|
2016-11-17 12:53:11 +00:00
|
|
|
nm_i->available_nids = nm_i->max_nid - sbi->total_valid_node_count -
|
2017-11-16 08:59:14 +00:00
|
|
|
sbi->nquota_files - F2FS_RESERVED_NODE_NUM;
|
2017-09-29 05:59:35 +00:00
|
|
|
nm_i->nid_cnt[FREE_NID] = 0;
|
|
|
|
nm_i->nid_cnt[PREALLOC_NID] = 0;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
nm_i->nat_cnt = 0;
|
2014-03-19 04:31:37 +00:00
|
|
|
nm_i->ram_thresh = DEF_RAM_THRESHOLD;
|
2015-10-12 09:08:48 +00:00
|
|
|
nm_i->ra_nid_pages = DEF_RA_NID_PAGES;
|
2016-01-18 10:32:58 +00:00
|
|
|
nm_i->dirty_nats_ratio = DEF_DIRTY_NAT_RATIO_THRESHOLD;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
2014-02-21 05:29:35 +00:00
|
|
|
INIT_RADIX_TREE(&nm_i->free_nid_root, GFP_ATOMIC);
|
2017-09-29 05:59:35 +00:00
|
|
|
INIT_LIST_HEAD(&nm_i->free_nid_list);
|
2014-12-04 04:47:26 +00:00
|
|
|
INIT_RADIX_TREE(&nm_i->nat_root, GFP_NOIO);
|
|
|
|
INIT_RADIX_TREE(&nm_i->nat_set_root, GFP_NOIO);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
INIT_LIST_HEAD(&nm_i->nat_entries);
|
|
|
|
|
|
|
|
mutex_init(&nm_i->build_lock);
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_lock_init(&nm_i->nid_list_lock);
|
2016-08-04 18:38:25 +00:00
|
|
|
init_rwsem(&nm_i->nat_tree_lock);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
nm_i->next_scan_nid = le32_to_cpu(sbi->ckpt->next_free_nid);
|
2013-03-28 00:24:53 +00:00
|
|
|
nm_i->bitmap_size = __bitmap_size(sbi, NAT_BITMAP);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
version_bitmap = __bitmap_ptr(sbi, NAT_BITMAP);
|
|
|
|
if (!version_bitmap)
|
|
|
|
return -EFAULT;
|
|
|
|
|
2013-03-28 00:24:53 +00:00
|
|
|
nm_i->nat_bitmap = kmemdup(version_bitmap, nm_i->bitmap_size,
|
|
|
|
GFP_KERNEL);
|
|
|
|
if (!nm_i->nat_bitmap)
|
|
|
|
return -ENOMEM;
|
2017-01-07 10:52:01 +00:00
|
|
|
|
2017-02-09 18:38:09 +00:00
|
|
|
err = __get_nat_bitmaps(sbi);
|
|
|
|
if (err)
|
|
|
|
return err;
|
|
|
|
|
2017-01-07 10:52:01 +00:00
|
|
|
#ifdef CONFIG_F2FS_CHECK_FS
|
|
|
|
nm_i->nat_bitmap_mir = kmemdup(version_bitmap, nm_i->bitmap_size,
|
|
|
|
GFP_KERNEL);
|
|
|
|
if (!nm_i->nat_bitmap_mir)
|
|
|
|
return -ENOMEM;
|
|
|
|
#endif
|
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2017-03-10 17:39:57 +00:00
|
|
|
static int init_free_nid_cache(struct f2fs_sb_info *sbi)
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
2018-03-10 01:42:28 +00:00
|
|
|
int i;
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
|
2018-03-10 01:42:28 +00:00
|
|
|
nm_i->free_nid_bitmap = f2fs_kzalloc(sbi, nm_i->nat_blocks *
|
|
|
|
sizeof(unsigned char *), GFP_KERNEL);
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
if (!nm_i->free_nid_bitmap)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
2018-03-10 01:42:28 +00:00
|
|
|
for (i = 0; i < nm_i->nat_blocks; i++) {
|
|
|
|
nm_i->free_nid_bitmap[i] = f2fs_kvzalloc(sbi,
|
|
|
|
NAT_ENTRY_BITMAP_SIZE_ALIGNED, GFP_KERNEL);
|
|
|
|
if (!nm_i->free_nid_bitmap)
|
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
|
2017-11-30 11:28:18 +00:00
|
|
|
nm_i->nat_block_bitmap = f2fs_kvzalloc(sbi, nm_i->nat_blocks / 8,
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
GFP_KERNEL);
|
|
|
|
if (!nm_i->nat_block_bitmap)
|
|
|
|
return -ENOMEM;
|
2017-03-01 09:09:07 +00:00
|
|
|
|
2017-11-30 11:28:18 +00:00
|
|
|
nm_i->free_nid_count = f2fs_kvzalloc(sbi, nm_i->nat_blocks *
|
2017-03-01 09:09:07 +00:00
|
|
|
sizeof(unsigned short), GFP_KERNEL);
|
|
|
|
if (!nm_i->free_nid_count)
|
|
|
|
return -ENOMEM;
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
int build_node_manager(struct f2fs_sb_info *sbi)
|
|
|
|
{
|
|
|
|
int err;
|
|
|
|
|
2017-11-30 11:28:17 +00:00
|
|
|
sbi->nm_info = f2fs_kzalloc(sbi, sizeof(struct f2fs_nm_info),
|
|
|
|
GFP_KERNEL);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (!sbi->nm_info)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
err = init_node_manager(sbi);
|
|
|
|
if (err)
|
|
|
|
return err;
|
|
|
|
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
err = init_free_nid_cache(sbi);
|
|
|
|
if (err)
|
|
|
|
return err;
|
|
|
|
|
2017-03-08 12:07:49 +00:00
|
|
|
/* load free nid status from nat_bits table */
|
|
|
|
load_free_nid_bitmap(sbi);
|
|
|
|
|
2017-02-09 18:38:09 +00:00
|
|
|
build_free_nids(sbi, true, true);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
void destroy_node_manager(struct f2fs_sb_info *sbi)
|
|
|
|
{
|
|
|
|
struct f2fs_nm_info *nm_i = NM_I(sbi);
|
|
|
|
struct free_nid *i, *next_i;
|
|
|
|
struct nat_entry *natvec[NATVEC_SIZE];
|
2015-01-07 18:47:57 +00:00
|
|
|
struct nat_entry_set *setvec[SETVEC_SIZE];
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
nid_t nid = 0;
|
|
|
|
unsigned int found;
|
|
|
|
|
|
|
|
if (!nm_i)
|
|
|
|
return;
|
|
|
|
|
|
|
|
/* destroy free nid list */
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_lock(&nm_i->nid_list_lock);
|
2017-09-29 05:59:35 +00:00
|
|
|
list_for_each_entry_safe(i, next_i, &nm_i->free_nid_list, list) {
|
2017-10-28 11:03:37 +00:00
|
|
|
__remove_free_nid(sbi, i, FREE_NID);
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_unlock(&nm_i->nid_list_lock);
|
2014-04-02 00:55:00 +00:00
|
|
|
kmem_cache_free(free_nid_slab, i);
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_lock(&nm_i->nid_list_lock);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
2017-09-29 05:59:35 +00:00
|
|
|
f2fs_bug_on(sbi, nm_i->nid_cnt[FREE_NID]);
|
|
|
|
f2fs_bug_on(sbi, nm_i->nid_cnt[PREALLOC_NID]);
|
|
|
|
f2fs_bug_on(sbi, !list_empty(&nm_i->free_nid_list));
|
f2fs: split free nid list
During free nid allocation, in order to do preallocation, we will tag free
nid entry as allocated one and still leave it in free nid list, for other
allocators who want to grab free nids, it needs to traverse the free nid
list for lookup. It becomes overhead in scenario of allocating free nid
intensively by multithreads.
This patch splits free nid list to two list: {free,alloc}_nid_list, to
keep free nids and preallocated free nids separately, after that, traverse
latency will be gone, besides split nid_cnt for separate statistic.
Additionally, introduce __insert_nid_to_list and __remove_nid_from_list for
cleanup.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: modify f2fs_bug_on to avoid needless branches]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-10-12 11:28:29 +00:00
|
|
|
spin_unlock(&nm_i->nid_list_lock);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
/* destroy nat cache */
|
2016-08-04 18:38:25 +00:00
|
|
|
down_write(&nm_i->nat_tree_lock);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
while ((found = __gang_lookup_nat_cache(nm_i,
|
|
|
|
nid, NATVEC_SIZE, natvec))) {
|
|
|
|
unsigned idx;
|
2015-01-07 18:47:57 +00:00
|
|
|
|
2014-03-07 10:43:24 +00:00
|
|
|
nid = nat_get_nid(natvec[found - 1]) + 1;
|
|
|
|
for (idx = 0; idx < found; idx++)
|
|
|
|
__del_from_nat_cache(nm_i, natvec[idx]);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
2014-09-02 22:52:58 +00:00
|
|
|
f2fs_bug_on(sbi, nm_i->nat_cnt);
|
2015-01-07 18:47:57 +00:00
|
|
|
|
|
|
|
/* destroy nat set cache */
|
|
|
|
nid = 0;
|
|
|
|
while ((found = __gang_lookup_nat_set(nm_i,
|
|
|
|
nid, SETVEC_SIZE, setvec))) {
|
|
|
|
unsigned idx;
|
|
|
|
|
|
|
|
nid = setvec[found - 1]->set + 1;
|
|
|
|
for (idx = 0; idx < found; idx++) {
|
|
|
|
/* entry_cnt is not zero, when cp_error was occurred */
|
|
|
|
f2fs_bug_on(sbi, !list_empty(&setvec[idx]->entry_list));
|
|
|
|
radix_tree_delete(&nm_i->nat_set_root, setvec[idx]->set);
|
|
|
|
kmem_cache_free(nat_entry_set_slab, setvec[idx]);
|
|
|
|
}
|
|
|
|
}
|
2016-08-04 18:38:25 +00:00
|
|
|
up_write(&nm_i->nat_tree_lock);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
kvfree(nm_i->nat_block_bitmap);
|
2018-03-10 01:42:28 +00:00
|
|
|
if (nm_i->free_nid_bitmap) {
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < nm_i->nat_blocks; i++)
|
|
|
|
kvfree(nm_i->free_nid_bitmap[i]);
|
|
|
|
kfree(nm_i->free_nid_bitmap);
|
|
|
|
}
|
2017-03-01 09:09:07 +00:00
|
|
|
kvfree(nm_i->free_nid_count);
|
f2fs: introduce free nid bitmap
In scenario of intensively node allocation, free nids will be ran out
soon, then it needs to stop to load free nids by traversing NAT blocks,
in worse case, if NAT blocks does not be cached in memory, it generates
IOs which slows down our foreground operations.
In order to speed up node allocation, in this patch we introduce a new
free_nid_bitmap array, so there is an bitmap table for each NAT block,
Once the NAT block is loaded, related bitmap cache will be switched on,
and bitmap will be set during traversing nat entries in NAT block, later
we can query and update nid usage status in memory completely.
With such implementation, I expect performance of node allocation can be
improved in the long-term after filesystem image is mounted.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-02-23 02:53:49 +00:00
|
|
|
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
kfree(nm_i->nat_bitmap);
|
2017-02-09 18:38:09 +00:00
|
|
|
kfree(nm_i->nat_bits);
|
2017-01-07 10:52:01 +00:00
|
|
|
#ifdef CONFIG_F2FS_CHECK_FS
|
|
|
|
kfree(nm_i->nat_bitmap_mir);
|
|
|
|
#endif
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
sbi->nm_info = NULL;
|
|
|
|
kfree(nm_i);
|
|
|
|
}
|
|
|
|
|
2013-01-16 15:08:30 +00:00
|
|
|
int __init create_node_manager_caches(void)
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
{
|
|
|
|
nat_entry_slab = f2fs_kmem_cache_create("nat_entry",
|
2014-03-07 10:43:28 +00:00
|
|
|
sizeof(struct nat_entry));
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
if (!nat_entry_slab)
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
goto fail;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
|
|
|
|
free_nid_slab = f2fs_kmem_cache_create("free_nid",
|
2014-03-07 10:43:28 +00:00
|
|
|
sizeof(struct free_nid));
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
if (!free_nid_slab)
|
2014-11-24 14:52:00 +00:00
|
|
|
goto destroy_nat_entry;
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
|
|
|
|
nat_entry_set_slab = f2fs_kmem_cache_create("nat_entry_set",
|
|
|
|
sizeof(struct nat_entry_set));
|
|
|
|
if (!nat_entry_set_slab)
|
2014-11-24 14:52:00 +00:00
|
|
|
goto destroy_free_nid;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
return 0;
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
|
2014-11-24 14:52:00 +00:00
|
|
|
destroy_free_nid:
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
kmem_cache_destroy(free_nid_slab);
|
2014-11-24 14:52:00 +00:00
|
|
|
destroy_nat_entry:
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
kmem_cache_destroy(nat_entry_slab);
|
|
|
|
fail:
|
|
|
|
return -ENOMEM;
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
void destroy_node_manager_caches(void)
|
|
|
|
{
|
f2fs: refactor flush_nat_entries codes for reducing NAT writes
Although building NAT journal in cursum reduce the read/write work for NAT
block, but previous design leave us lower performance when write checkpoint
frequently for these cases:
1. if journal in cursum has already full, it's a bit of waste that we flush all
nat entries to page for persistence, but not to cache any entries.
2. if journal in cursum is not full, we fill nat entries to journal util
journal is full, then flush the left dirty entries to disk without merge
journaled entries, so these journaled entries may be flushed to disk at next
checkpoint but lost chance to flushed last time.
In this patch we merge dirty entries located in same NAT block to nat entry set,
and linked all set to list, sorted ascending order by entries' count of set.
Later we flush entries in sparse set into journal as many as we can, and then
flush merged entries to disk. In this way we can not only gain in performance,
but also save lifetime of flash device.
In my testing environment, it shows this patch can help to reduce NAT block
writes obviously. In hard disk test case: cost time of fsstress is stablely
reduced by about 5%.
1. virtual machine + hard disk:
fsstress -p 20 -n 200 -l 5
node num cp count nodes/cp
based 4599.6 1803.0 2.551
patched 2714.6 1829.6 1.483
2. virtual machine + 32g micro SD card:
fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0
-f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5
-f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S
node num cp count nodes/cp
based 84.5 43.7 1.933
patched 49.2 40.0 1.23
Our latency of merging op shows not bad when handling extreme case like:
merging a great number of dirty nats:
latency(ns) dirty nat count
3089219 24922
5129423 27422
4000250 24523
change log from v1:
o fix wrong logic in add_nat_entry when grab a new nat entry set.
o swith to create slab cache in create_node_manager_caches.
o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency.
change log from v2:
o make comment position more appropriate suggested by Jaegeuk Kim.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-06-24 01:18:20 +00:00
|
|
|
kmem_cache_destroy(nat_entry_set_slab);
|
f2fs: add node operations
This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.
- The key information of an NAT entry consists of a node id and a block address.
- An NAT page is composed of block addresses covered by a certain range of NAT
entries, which is maintained by the address space of meta_inode.
- A radix tree structure is used to cache NAT entries. The index for the tree
is a node id.
- When there is no free nid, F2FS should scan NAT entries to find new one. In
order to avoid scanning frequently, F2FS manages a list containing a number of
free nids in memory. Only when free nids in the list are exhausted, scanning
process, build_free_nids(), is triggered.
- F2FS has direct and indirect node blocks for indexing data. This patch adds
fuctions related to the node block management such as getting, allocating, and
truncating node blocks to index data.
- In order to cache node blocks in memory, F2FS has a node_inode with an address
space for node pages. This patch also adds the address space operations for
node_inode.
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-02 08:08:50 +00:00
|
|
|
kmem_cache_destroy(free_nid_slab);
|
|
|
|
kmem_cache_destroy(nat_entry_slab);
|
|
|
|
}
|