Previously, f2fs postpones reclaiming prefree segments into free segments
as much as possible.
However, if user writes and deletes a bunch of data without any sync or fsync
calls, some flash storages can suffer from garbage collections.
So, this patch adds the reclaiming codes to f2fs_write_node_pages and background
GC thread.
If there are a lot of prefree segments, let's do checkpoint so that f2fs
submits discard commands for the prefree regions to the flash storage.
Signed-off-by: Changman Lee <[email protected]>
Signed-off-by: Jaegeuk Kim <[email protected]>
---
fs/f2fs/f2fs.h | 3 +++
fs/f2fs/gc.c | 5 +++++
fs/f2fs/node.c | 3 ++-
fs/f2fs/segment.c | 1 +
fs/f2fs/segment.h | 7 +++++++
5 files changed, 18 insertions(+), 1 deletion(-)
diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
index 590a09e..3c95d35 100644
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -299,6 +299,9 @@ struct f2fs_sm_info {
unsigned int main_segments; /* # of segments in main area */
unsigned int reserved_segments; /* # of reserved segments */
unsigned int ovp_segments; /* # of overprovision segments */
+
+ /* a threshold to reclaim prefree segments */
+ unsigned int rec_prefree_segments;
};
/*
diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c
index cb286d7..783c6cc 100644
--- a/fs/f2fs/gc.c
+++ b/fs/f2fs/gc.c
@@ -82,6 +82,11 @@ static int gc_thread_func(void *data)
/* if return value is not zero, no victim was selected */
if (f2fs_gc(sbi))
wait_ms = gc_th->no_gc_sleep_time;
+
+ /* balancing prefree segments */
+ if (excess_prefree_segs(sbi))
+ f2fs_sync_fs(sbi->sb, true);
+
} while (!kthread_should_stop());
return 0;
}
diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
index cc119b6..89dd8a5 100644
--- a/fs/f2fs/node.c
+++ b/fs/f2fs/node.c
@@ -1206,7 +1206,8 @@ static int f2fs_write_node_pages(struct address_space *mapping,
long nr_to_write = wbc->nr_to_write;
/* First check balancing cached NAT entries */
- if (try_to_free_nats(sbi, NAT_ENTRY_PER_BLOCK)) {
+ if (try_to_free_nats(sbi, NAT_ENTRY_PER_BLOCK) ||
+ excess_prefree_segs(sbi)) {
f2fs_sync_fs(sbi->sb, true);
return 0;
}
diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index c9c276e..43ee91e 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -1645,6 +1645,7 @@ int build_segment_manager(struct f2fs_sb_info *sbi)
sm_info->ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
sm_info->main_segments = le32_to_cpu(raw_super->segment_count_main);
sm_info->ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
+ sm_info->rec_prefree_segments = DEF_RECLAIM_PREFREE_SEGMENTS;
err = build_sit_info(sbi);
if (err)
diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h
index 7f94d78..abe7094 100644
--- a/fs/f2fs/segment.h
+++ b/fs/f2fs/segment.h
@@ -14,6 +14,8 @@
#define NULL_SEGNO ((unsigned int)(~0))
#define NULL_SECNO ((unsigned int)(~0))
+#define DEF_RECLAIM_PREFREE_SEGMENTS 100 /* 200MB of prefree segments */
+
/* L: Logical segment # in volume, R: Relative segment # in main area */
#define GET_L2R_SEGNO(free_i, segno) (segno - free_i->start_segno)
#define GET_R2L_SEGNO(free_i, segno) (segno + free_i->start_segno)
@@ -472,6 +474,11 @@ static inline bool has_not_enough_free_secs(struct f2fs_sb_info *sbi, int freed)
reserved_sections(sbi)));
}
+static inline bool excess_prefree_segs(struct f2fs_sb_info *sbi)
+{
+ return (prefree_segments(sbi) > SM_I(sbi)->rec_prefree_segments);
+}
+
static inline int utilization(struct f2fs_sb_info *sbi)
{
return div_u64((u64)valid_user_blocks(sbi) * 100, sbi->user_block_count);
--
1.8.4.474.g128a96c
This patch merges some background jobs into this new function.
Signed-off-by: Changman Lee <[email protected]>
Signed-off-by: Jaegeuk Kim <[email protected]>
---
fs/f2fs/f2fs.h | 2 ++
fs/f2fs/gc.c | 5 ++---
fs/f2fs/node.c | 10 +++-------
fs/f2fs/segment.c | 8 ++++++++
4 files changed, 15 insertions(+), 10 deletions(-)
diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
index 3c95d35..602a41d 100644
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -965,6 +965,7 @@ long f2fs_compat_ioctl(struct file *, unsigned int, unsigned long);
*/
void f2fs_set_inode_flags(struct inode *);
struct inode *f2fs_iget(struct super_block *, unsigned long);
+int try_to_free_nats(struct f2fs_sb_info *, int);
void update_inode(struct inode *, struct page *);
int update_inode_page(struct inode *);
int f2fs_write_inode(struct inode *, struct writeback_control *);
@@ -1045,6 +1046,7 @@ void destroy_node_manager_caches(void);
* segment.c
*/
void f2fs_balance_fs(struct f2fs_sb_info *);
+void f2fs_balance_fs_bg(struct f2fs_sb_info *);
void invalidate_blocks(struct f2fs_sb_info *, block_t);
void clear_prefree_segments(struct f2fs_sb_info *);
int npages_for_summary_flush(struct f2fs_sb_info *);
diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c
index 783c6cc..b7ad1ec 100644
--- a/fs/f2fs/gc.c
+++ b/fs/f2fs/gc.c
@@ -83,9 +83,8 @@ static int gc_thread_func(void *data)
if (f2fs_gc(sbi))
wait_ms = gc_th->no_gc_sleep_time;
- /* balancing prefree segments */
- if (excess_prefree_segs(sbi))
- f2fs_sync_fs(sbi->sb, true);
+ /* balancing f2fs's metadata periodically */
+ f2fs_balance_fs_bg(sbi);
} while (!kthread_should_stop());
return 0;
diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
index 89dd8a5..7bac481 100644
--- a/fs/f2fs/node.c
+++ b/fs/f2fs/node.c
@@ -240,7 +240,7 @@ retry:
write_unlock(&nm_i->nat_tree_lock);
}
-static int try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink)
+int try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink)
{
struct f2fs_nm_info *nm_i = NM_I(sbi);
@@ -1205,12 +1205,8 @@ static int f2fs_write_node_pages(struct address_space *mapping,
struct f2fs_sb_info *sbi = F2FS_SB(mapping->host->i_sb);
long nr_to_write = wbc->nr_to_write;
- /* First check balancing cached NAT entries */
- if (try_to_free_nats(sbi, NAT_ENTRY_PER_BLOCK) ||
- excess_prefree_segs(sbi)) {
- f2fs_sync_fs(sbi->sb, true);
- return 0;
- }
+ /* balancing f2fs's metadata in background */
+ f2fs_balance_fs_bg(sbi);
/* collect a number of dirty node pages and write together */
if (get_pages(sbi, F2FS_DIRTY_NODES) < COLLECT_DIRTY_NODES)
diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index 43ee91e..77aee83 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -36,6 +36,14 @@ void f2fs_balance_fs(struct f2fs_sb_info *sbi)
}
}
+void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi)
+{
+ /* check the # of cached NAT entries and prefree segments */
+ if (try_to_free_nats(sbi, NAT_ENTRY_PER_BLOCK) ||
+ excess_prefree_segs(sbi))
+ f2fs_sync_fs(sbi->sb, true);
+}
+
static void __locate_dirty_segment(struct f2fs_sb_info *sbi, unsigned int segno,
enum dirty_type dirty_type)
{
--
1.8.4.474.g128a96c
This patch adds a control method in sysfs to reclaim prefree segments.
Signed-off-by: Changman Lee <[email protected]>
Signed-off-by: Jaegeuk Kim <[email protected]>
---
Documentation/filesystems/f2fs.txt | 6 +++++
fs/f2fs/super.c | 50 +++++++++++++++++++++++++++-----------
2 files changed, 42 insertions(+), 14 deletions(-)
diff --git a/Documentation/filesystems/f2fs.txt b/Documentation/filesystems/f2fs.txt
index 3cd27be..4c647c2 100644
--- a/Documentation/filesystems/f2fs.txt
+++ b/Documentation/filesystems/f2fs.txt
@@ -164,6 +164,12 @@ Files in /sys/fs/f2fs/<devname>
gc_idle = 1 will select the Cost Benefit approach
& setting gc_idle = 2 will select the greedy aproach.
+ reclaim_segments This parameter controls the number of prefree
+ segments to be reclaimed. If the number of prefree
+ segments is larger than this number, f2fs tries to
+ conduct checkpoint to reclaim the prefree segments
+ to free segments. By default, 100 segments, 200MB.
+
================================================================================
USAGE
================================================================================
diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c
index 9a09459..e42351c 100644
--- a/fs/f2fs/super.c
+++ b/fs/f2fs/super.c
@@ -69,24 +69,40 @@ static match_table_t f2fs_tokens = {
};
/* Sysfs support for f2fs */
+enum {
+ GC_THREAD, /* struct f2fs_gc_thread */
+ SM_INFO, /* struct f2fs_sm_info */
+};
+
struct f2fs_attr {
struct attribute attr;
ssize_t (*show)(struct f2fs_attr *, struct f2fs_sb_info *, char *);
ssize_t (*store)(struct f2fs_attr *, struct f2fs_sb_info *,
const char *, size_t);
+ int struct_type;
int offset;
};
+static unsigned char *__struct_ptr(struct f2fs_sb_info *sbi, int struct_type)
+{
+ if (struct_type == GC_THREAD)
+ return (unsigned char *)sbi->gc_thread;
+ else if (struct_type == SM_INFO)
+ return (unsigned char *)SM_I(sbi);
+ return NULL;
+}
+
static ssize_t f2fs_sbi_show(struct f2fs_attr *a,
struct f2fs_sb_info *sbi, char *buf)
{
- struct f2fs_gc_kthread *gc_kth = sbi->gc_thread;
+ unsigned char *ptr = NULL;
unsigned int *ui;
- if (!gc_kth)
+ ptr = __struct_ptr(sbi, a->struct_type);
+ if (!ptr)
return -EINVAL;
- ui = (unsigned int *)(((char *)gc_kth) + a->offset);
+ ui = (unsigned int *)(ptr + a->offset);
return snprintf(buf, PAGE_SIZE, "%u\n", *ui);
}
@@ -95,15 +111,16 @@ static ssize_t f2fs_sbi_store(struct f2fs_attr *a,
struct f2fs_sb_info *sbi,
const char *buf, size_t count)
{
- struct f2fs_gc_kthread *gc_kth = sbi->gc_thread;
+ unsigned char *ptr;
unsigned long t;
unsigned int *ui;
ssize_t ret;
- if (!gc_kth)
+ ptr = __struct_ptr(sbi, a->struct_type);
+ if (!ptr)
return -EINVAL;
- ui = (unsigned int *)(((char *)gc_kth) + a->offset);
+ ui = (unsigned int *)(ptr + a->offset);
ret = kstrtoul(skip_spaces(buf), 0, &t);
if (ret < 0)
@@ -139,21 +156,25 @@ static void f2fs_sb_release(struct kobject *kobj)
complete(&sbi->s_kobj_unregister);
}
-#define F2FS_ATTR_OFFSET(_name, _mode, _show, _store, _elname) \
+#define F2FS_ATTR_OFFSET(_struct_type, _name, _mode, _show, _store, _offset) \
static struct f2fs_attr f2fs_attr_##_name = { \
.attr = {.name = __stringify(_name), .mode = _mode }, \
.show = _show, \
.store = _store, \
- .offset = offsetof(struct f2fs_gc_kthread, _elname), \
+ .struct_type = _struct_type, \
+ .offset = _offset \
}
-#define F2FS_RW_ATTR(name, elname) \
- F2FS_ATTR_OFFSET(name, 0644, f2fs_sbi_show, f2fs_sbi_store, elname)
+#define F2FS_RW_ATTR(struct_type, struct_name, name, elname) \
+ F2FS_ATTR_OFFSET(struct_type, name, 0644, \
+ f2fs_sbi_show, f2fs_sbi_store, \
+ offsetof(struct struct_name, elname))
-F2FS_RW_ATTR(gc_min_sleep_time, min_sleep_time);
-F2FS_RW_ATTR(gc_max_sleep_time, max_sleep_time);
-F2FS_RW_ATTR(gc_no_gc_sleep_time, no_gc_sleep_time);
-F2FS_RW_ATTR(gc_idle, gc_idle);
+F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_min_sleep_time, min_sleep_time);
+F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_max_sleep_time, max_sleep_time);
+F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_no_gc_sleep_time, no_gc_sleep_time);
+F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_idle, gc_idle);
+F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, reclaim_segments, rec_prefree_segments);
#define ATTR_LIST(name) (&f2fs_attr_##name.attr)
static struct attribute *f2fs_attrs[] = {
@@ -161,6 +182,7 @@ static struct attribute *f2fs_attrs[] = {
ATTR_LIST(gc_max_sleep_time),
ATTR_LIST(gc_no_gc_sleep_time),
ATTR_LIST(gc_idle),
+ ATTR_LIST(reclaim_segments),
NULL,
};
--
1.8.4.474.g128a96c