Update Linux to v5.10.109
Sourced from [1]
[1] https://cdn.kernel.org/pub/linux/kernel/v5.x/linux-5.10.109.tar.xz
Change-Id: I19bca9fc6762d4e63bcf3e4cba88bbe560d9c76c
Signed-off-by: Olivier Deprez <olivier.deprez@arm.com>
diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c
index 4b6c362..6b240b7 100644
--- a/fs/f2fs/gc.c
+++ b/fs/f2fs/gc.c
@@ -13,6 +13,7 @@
#include <linux/kthread.h>
#include <linux/delay.h>
#include <linux/freezer.h>
+#include <linux/sched/signal.h>
#include "f2fs.h"
#include "node.h"
@@ -20,6 +21,11 @@
#include "gc.h"
#include <trace/events/f2fs.h>
+static struct kmem_cache *victim_entry_slab;
+
+static unsigned int count_bits(const unsigned long *addr,
+ unsigned int offset, unsigned int len);
+
static int gc_thread_func(void *data)
{
struct f2fs_sb_info *sbi = data;
@@ -31,6 +37,8 @@ static int gc_thread_func(void *data)
set_freezable();
do {
+ bool sync_mode;
+
wait_event_interruptible_timeout(*wq,
kthread_should_stop() || freezing(current) ||
gc_th->gc_wake,
@@ -76,20 +84,20 @@ static int gc_thread_func(void *data)
* invalidated soon after by user update or deletion.
* So, I'd like to wait some time to collect dirty segments.
*/
- if (sbi->gc_mode == GC_URGENT) {
+ if (sbi->gc_mode == GC_URGENT_HIGH) {
wait_ms = gc_th->urgent_sleep_time;
- mutex_lock(&sbi->gc_mutex);
+ down_write(&sbi->gc_lock);
goto do_gc;
}
- if (!mutex_trylock(&sbi->gc_mutex)) {
+ if (!down_write_trylock(&sbi->gc_lock)) {
stat_other_skip_bggc_count(sbi);
goto next;
}
if (!is_idle(sbi, GC_TIME)) {
increase_sleep_time(gc_th, &wait_ms);
- mutex_unlock(&sbi->gc_mutex);
+ up_write(&sbi->gc_lock);
stat_io_skip_bggc_count(sbi);
goto next;
}
@@ -99,17 +107,19 @@ static int gc_thread_func(void *data)
else
increase_sleep_time(gc_th, &wait_ms);
do_gc:
- stat_inc_bggc_count(sbi);
+ stat_inc_bggc_count(sbi->stat_info);
+
+ sync_mode = F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC;
/* if return value is not zero, no victim was selected */
- if (f2fs_gc(sbi, test_opt(sbi, FORCE_FG_GC), true, NULL_SEGNO))
+ if (f2fs_gc(sbi, sync_mode, true, false, NULL_SEGNO))
wait_ms = gc_th->no_gc_sleep_time;
trace_f2fs_background_gc(sbi->sb, wait_ms,
prefree_segments(sbi), free_segments(sbi));
/* balancing f2fs's metadata periodically */
- f2fs_balance_fs_bg(sbi);
+ f2fs_balance_fs_bg(sbi, true);
next:
sb_end_write(sbi->sb);
@@ -142,7 +152,7 @@ int f2fs_start_gc_thread(struct f2fs_sb_info *sbi)
"f2fs_gc-%u:%u", MAJOR(dev), MINOR(dev));
if (IS_ERR(gc_th->f2fs_gc_task)) {
err = PTR_ERR(gc_th->f2fs_gc_task);
- kvfree(gc_th);
+ kfree(gc_th);
sbi->gc_thread = NULL;
}
out:
@@ -155,23 +165,36 @@ void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi)
if (!gc_th)
return;
kthread_stop(gc_th->f2fs_gc_task);
- kvfree(gc_th);
+ kfree(gc_th);
sbi->gc_thread = NULL;
}
static int select_gc_type(struct f2fs_sb_info *sbi, int gc_type)
{
- int gc_mode = (gc_type == BG_GC) ? GC_CB : GC_GREEDY;
+ int gc_mode;
+
+ if (gc_type == BG_GC) {
+ if (sbi->am.atgc_enabled)
+ gc_mode = GC_AT;
+ else
+ gc_mode = GC_CB;
+ } else {
+ gc_mode = GC_GREEDY;
+ }
switch (sbi->gc_mode) {
case GC_IDLE_CB:
gc_mode = GC_CB;
break;
case GC_IDLE_GREEDY:
- case GC_URGENT:
+ case GC_URGENT_HIGH:
gc_mode = GC_GREEDY;
break;
+ case GC_IDLE_AT:
+ gc_mode = GC_AT;
+ break;
}
+
return gc_mode;
}
@@ -182,19 +205,34 @@ static void select_policy(struct f2fs_sb_info *sbi, int gc_type,
if (p->alloc_mode == SSR) {
p->gc_mode = GC_GREEDY;
- p->dirty_segmap = dirty_i->dirty_segmap[type];
+ p->dirty_bitmap = dirty_i->dirty_segmap[type];
+ p->max_search = dirty_i->nr_dirty[type];
+ p->ofs_unit = 1;
+ } else if (p->alloc_mode == AT_SSR) {
+ p->gc_mode = GC_GREEDY;
+ p->dirty_bitmap = dirty_i->dirty_segmap[type];
p->max_search = dirty_i->nr_dirty[type];
p->ofs_unit = 1;
} else {
p->gc_mode = select_gc_type(sbi, gc_type);
- p->dirty_segmap = dirty_i->dirty_segmap[DIRTY];
- p->max_search = dirty_i->nr_dirty[DIRTY];
p->ofs_unit = sbi->segs_per_sec;
+ if (__is_large_section(sbi)) {
+ p->dirty_bitmap = dirty_i->dirty_secmap;
+ p->max_search = count_bits(p->dirty_bitmap,
+ 0, MAIN_SECS(sbi));
+ } else {
+ p->dirty_bitmap = dirty_i->dirty_segmap[DIRTY];
+ p->max_search = dirty_i->nr_dirty[DIRTY];
+ }
}
- /* we need to check every dirty segments in the FG_GC case */
+ /*
+ * adjust candidates range, should select all dirty segments for
+ * foreground GC and urgent GC cases.
+ */
if (gc_type != FG_GC &&
- (sbi->gc_mode != GC_URGENT) &&
+ (sbi->gc_mode != GC_URGENT_HIGH) &&
+ (p->gc_mode != GC_AT && p->alloc_mode != AT_SSR) &&
p->max_search > sbi->max_victim_search)
p->max_search = sbi->max_victim_search;
@@ -212,10 +250,16 @@ static unsigned int get_max_cost(struct f2fs_sb_info *sbi,
/* SSR allocates in a segment unit */
if (p->alloc_mode == SSR)
return sbi->blocks_per_seg;
+ else if (p->alloc_mode == AT_SSR)
+ return UINT_MAX;
+
+ /* LFS */
if (p->gc_mode == GC_GREEDY)
return 2 * sbi->blocks_per_seg * p->ofs_unit;
else if (p->gc_mode == GC_CB)
return UINT_MAX;
+ else if (p->gc_mode == GC_AT)
+ return UINT_MAX;
else /* No other gc_mode */
return 0;
}
@@ -249,13 +293,14 @@ static unsigned int get_cb_cost(struct f2fs_sb_info *sbi, unsigned int segno)
unsigned char age = 0;
unsigned char u;
unsigned int i;
+ unsigned int usable_segs_per_sec = f2fs_usable_segs_in_sec(sbi, segno);
- for (i = 0; i < sbi->segs_per_sec; i++)
+ for (i = 0; i < usable_segs_per_sec; i++)
mtime += get_seg_entry(sbi, start + i)->mtime;
vblocks = get_valid_blocks(sbi, segno, true);
- mtime = div_u64(mtime, sbi->segs_per_sec);
- vblocks = div_u64(vblocks, sbi->segs_per_sec);
+ mtime = div_u64(mtime, usable_segs_per_sec);
+ vblocks = div_u64(vblocks, usable_segs_per_sec);
u = (vblocks * 100) >> sbi->log_blocks_per_seg;
@@ -280,8 +325,11 @@ static inline unsigned int get_gc_cost(struct f2fs_sb_info *sbi,
/* alloc_mode == LFS */
if (p->gc_mode == GC_GREEDY)
return get_valid_blocks(sbi, segno, true);
- else
+ else if (p->gc_mode == GC_CB)
return get_cb_cost(sbi, segno);
+
+ f2fs_bug_on(sbi, 1);
+ return 0;
}
static unsigned int count_bits(const unsigned long *addr,
@@ -296,6 +344,269 @@ static unsigned int count_bits(const unsigned long *addr,
return sum;
}
+static struct victim_entry *attach_victim_entry(struct f2fs_sb_info *sbi,
+ unsigned long long mtime, unsigned int segno,
+ struct rb_node *parent, struct rb_node **p,
+ bool left_most)
+{
+ struct atgc_management *am = &sbi->am;
+ struct victim_entry *ve;
+
+ ve = f2fs_kmem_cache_alloc(victim_entry_slab, GFP_NOFS);
+
+ ve->mtime = mtime;
+ ve->segno = segno;
+
+ rb_link_node(&ve->rb_node, parent, p);
+ rb_insert_color_cached(&ve->rb_node, &am->root, left_most);
+
+ list_add_tail(&ve->list, &am->victim_list);
+
+ am->victim_count++;
+
+ return ve;
+}
+
+static void insert_victim_entry(struct f2fs_sb_info *sbi,
+ unsigned long long mtime, unsigned int segno)
+{
+ struct atgc_management *am = &sbi->am;
+ struct rb_node **p;
+ struct rb_node *parent = NULL;
+ bool left_most = true;
+
+ p = f2fs_lookup_rb_tree_ext(sbi, &am->root, &parent, mtime, &left_most);
+ attach_victim_entry(sbi, mtime, segno, parent, p, left_most);
+}
+
+static void add_victim_entry(struct f2fs_sb_info *sbi,
+ struct victim_sel_policy *p, unsigned int segno)
+{
+ struct sit_info *sit_i = SIT_I(sbi);
+ unsigned int secno = GET_SEC_FROM_SEG(sbi, segno);
+ unsigned int start = GET_SEG_FROM_SEC(sbi, secno);
+ unsigned long long mtime = 0;
+ unsigned int i;
+
+ if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
+ if (p->gc_mode == GC_AT &&
+ get_valid_blocks(sbi, segno, true) == 0)
+ return;
+ }
+
+ for (i = 0; i < sbi->segs_per_sec; i++)
+ mtime += get_seg_entry(sbi, start + i)->mtime;
+ mtime = div_u64(mtime, sbi->segs_per_sec);
+
+ /* Handle if the system time has changed by the user */
+ if (mtime < sit_i->min_mtime)
+ sit_i->min_mtime = mtime;
+ if (mtime > sit_i->max_mtime)
+ sit_i->max_mtime = mtime;
+ if (mtime < sit_i->dirty_min_mtime)
+ sit_i->dirty_min_mtime = mtime;
+ if (mtime > sit_i->dirty_max_mtime)
+ sit_i->dirty_max_mtime = mtime;
+
+ /* don't choose young section as candidate */
+ if (sit_i->dirty_max_mtime - mtime < p->age_threshold)
+ return;
+
+ insert_victim_entry(sbi, mtime, segno);
+}
+
+static struct rb_node *lookup_central_victim(struct f2fs_sb_info *sbi,
+ struct victim_sel_policy *p)
+{
+ struct atgc_management *am = &sbi->am;
+ struct rb_node *parent = NULL;
+ bool left_most;
+
+ f2fs_lookup_rb_tree_ext(sbi, &am->root, &parent, p->age, &left_most);
+
+ return parent;
+}
+
+static void atgc_lookup_victim(struct f2fs_sb_info *sbi,
+ struct victim_sel_policy *p)
+{
+ struct sit_info *sit_i = SIT_I(sbi);
+ struct atgc_management *am = &sbi->am;
+ struct rb_root_cached *root = &am->root;
+ struct rb_node *node;
+ struct rb_entry *re;
+ struct victim_entry *ve;
+ unsigned long long total_time;
+ unsigned long long age, u, accu;
+ unsigned long long max_mtime = sit_i->dirty_max_mtime;
+ unsigned long long min_mtime = sit_i->dirty_min_mtime;
+ unsigned int sec_blocks = BLKS_PER_SEC(sbi);
+ unsigned int vblocks;
+ unsigned int dirty_threshold = max(am->max_candidate_count,
+ am->candidate_ratio *
+ am->victim_count / 100);
+ unsigned int age_weight = am->age_weight;
+ unsigned int cost;
+ unsigned int iter = 0;
+
+ if (max_mtime < min_mtime)
+ return;
+
+ max_mtime += 1;
+ total_time = max_mtime - min_mtime;
+
+ accu = div64_u64(ULLONG_MAX, total_time);
+ accu = min_t(unsigned long long, div_u64(accu, 100),
+ DEFAULT_ACCURACY_CLASS);
+
+ node = rb_first_cached(root);
+next:
+ re = rb_entry_safe(node, struct rb_entry, rb_node);
+ if (!re)
+ return;
+
+ ve = (struct victim_entry *)re;
+
+ if (ve->mtime >= max_mtime || ve->mtime < min_mtime)
+ goto skip;
+
+ /* age = 10000 * x% * 60 */
+ age = div64_u64(accu * (max_mtime - ve->mtime), total_time) *
+ age_weight;
+
+ vblocks = get_valid_blocks(sbi, ve->segno, true);
+ f2fs_bug_on(sbi, !vblocks || vblocks == sec_blocks);
+
+ /* u = 10000 * x% * 40 */
+ u = div64_u64(accu * (sec_blocks - vblocks), sec_blocks) *
+ (100 - age_weight);
+
+ f2fs_bug_on(sbi, age + u >= UINT_MAX);
+
+ cost = UINT_MAX - (age + u);
+ iter++;
+
+ if (cost < p->min_cost ||
+ (cost == p->min_cost && age > p->oldest_age)) {
+ p->min_cost = cost;
+ p->oldest_age = age;
+ p->min_segno = ve->segno;
+ }
+skip:
+ if (iter < dirty_threshold) {
+ node = rb_next(node);
+ goto next;
+ }
+}
+
+/*
+ * select candidates around source section in range of
+ * [target - dirty_threshold, target + dirty_threshold]
+ */
+static void atssr_lookup_victim(struct f2fs_sb_info *sbi,
+ struct victim_sel_policy *p)
+{
+ struct sit_info *sit_i = SIT_I(sbi);
+ struct atgc_management *am = &sbi->am;
+ struct rb_node *node;
+ struct rb_entry *re;
+ struct victim_entry *ve;
+ unsigned long long age;
+ unsigned long long max_mtime = sit_i->dirty_max_mtime;
+ unsigned long long min_mtime = sit_i->dirty_min_mtime;
+ unsigned int seg_blocks = sbi->blocks_per_seg;
+ unsigned int vblocks;
+ unsigned int dirty_threshold = max(am->max_candidate_count,
+ am->candidate_ratio *
+ am->victim_count / 100);
+ unsigned int cost;
+ unsigned int iter = 0;
+ int stage = 0;
+
+ if (max_mtime < min_mtime)
+ return;
+ max_mtime += 1;
+next_stage:
+ node = lookup_central_victim(sbi, p);
+next_node:
+ re = rb_entry_safe(node, struct rb_entry, rb_node);
+ if (!re) {
+ if (stage == 0)
+ goto skip_stage;
+ return;
+ }
+
+ ve = (struct victim_entry *)re;
+
+ if (ve->mtime >= max_mtime || ve->mtime < min_mtime)
+ goto skip_node;
+
+ age = max_mtime - ve->mtime;
+
+ vblocks = get_seg_entry(sbi, ve->segno)->ckpt_valid_blocks;
+ f2fs_bug_on(sbi, !vblocks);
+
+ /* rare case */
+ if (vblocks == seg_blocks)
+ goto skip_node;
+
+ iter++;
+
+ age = max_mtime - abs(p->age - age);
+ cost = UINT_MAX - vblocks;
+
+ if (cost < p->min_cost ||
+ (cost == p->min_cost && age > p->oldest_age)) {
+ p->min_cost = cost;
+ p->oldest_age = age;
+ p->min_segno = ve->segno;
+ }
+skip_node:
+ if (iter < dirty_threshold) {
+ if (stage == 0)
+ node = rb_prev(node);
+ else if (stage == 1)
+ node = rb_next(node);
+ goto next_node;
+ }
+skip_stage:
+ if (stage < 1) {
+ stage++;
+ iter = 0;
+ goto next_stage;
+ }
+}
+static void lookup_victim_by_age(struct f2fs_sb_info *sbi,
+ struct victim_sel_policy *p)
+{
+ f2fs_bug_on(sbi, !f2fs_check_rb_tree_consistence(sbi,
+ &sbi->am.root, true));
+
+ if (p->gc_mode == GC_AT)
+ atgc_lookup_victim(sbi, p);
+ else if (p->alloc_mode == AT_SSR)
+ atssr_lookup_victim(sbi, p);
+ else
+ f2fs_bug_on(sbi, 1);
+}
+
+static void release_victim_entry(struct f2fs_sb_info *sbi)
+{
+ struct atgc_management *am = &sbi->am;
+ struct victim_entry *ve, *tmp;
+
+ list_for_each_entry_safe(ve, tmp, &am->victim_list, list) {
+ list_del(&ve->list);
+ kmem_cache_free(victim_entry_slab, ve);
+ am->victim_count--;
+ }
+
+ am->root = RB_ROOT_CACHED;
+
+ f2fs_bug_on(sbi, am->victim_count);
+ f2fs_bug_on(sbi, !list_empty(&am->victim_list));
+}
+
/*
* This function is called from two paths.
* One is garbage collection and the other is SSR segment selection.
@@ -305,31 +616,51 @@ static unsigned int count_bits(const unsigned long *addr,
* which has minimum valid blocks and removes it from dirty seglist.
*/
static int get_victim_by_default(struct f2fs_sb_info *sbi,
- unsigned int *result, int gc_type, int type, char alloc_mode)
+ unsigned int *result, int gc_type, int type,
+ char alloc_mode, unsigned long long age)
{
struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
struct sit_info *sm = SIT_I(sbi);
struct victim_sel_policy p;
unsigned int secno, last_victim;
unsigned int last_segment;
- unsigned int nsearched = 0;
+ unsigned int nsearched;
+ bool is_atgc;
+ int ret = 0;
mutex_lock(&dirty_i->seglist_lock);
last_segment = MAIN_SECS(sbi) * sbi->segs_per_sec;
p.alloc_mode = alloc_mode;
- select_policy(sbi, gc_type, type, &p);
+ p.age = age;
+ p.age_threshold = sbi->am.age_threshold;
+retry:
+ select_policy(sbi, gc_type, type, &p);
p.min_segno = NULL_SEGNO;
+ p.oldest_age = 0;
p.min_cost = get_max_cost(sbi, &p);
+ is_atgc = (p.gc_mode == GC_AT || p.alloc_mode == AT_SSR);
+ nsearched = 0;
+
+ if (is_atgc)
+ SIT_I(sbi)->dirty_min_mtime = ULLONG_MAX;
+
if (*result != NULL_SEGNO) {
- if (get_valid_blocks(sbi, *result, false) &&
- !sec_usage_check(sbi, GET_SEC_FROM_SEG(sbi, *result)))
+ if (!get_valid_blocks(sbi, *result, false)) {
+ ret = -ENODATA;
+ goto out;
+ }
+
+ if (sec_usage_check(sbi, GET_SEC_FROM_SEG(sbi, *result)))
+ ret = -EBUSY;
+ else
p.min_segno = *result;
goto out;
}
+ ret = -ENODATA;
if (p.max_search == 0)
goto out;
@@ -357,10 +688,14 @@ static int get_victim_by_default(struct f2fs_sb_info *sbi,
}
while (1) {
- unsigned long cost;
- unsigned int segno;
+ unsigned long cost, *dirty_bitmap;
+ unsigned int unit_no, segno;
- segno = find_next_bit(p.dirty_segmap, last_segment, p.offset);
+ dirty_bitmap = p.dirty_bitmap;
+ unit_no = find_next_bit(dirty_bitmap,
+ last_segment / p.ofs_unit,
+ p.offset / p.ofs_unit);
+ segno = unit_no * p.ofs_unit;
if (segno >= last_segment) {
if (sm->last_victim[p.gc_mode]) {
last_segment =
@@ -373,14 +708,7 @@ static int get_victim_by_default(struct f2fs_sb_info *sbi,
}
p.offset = segno + p.ofs_unit;
- if (p.ofs_unit > 1) {
- p.offset -= segno % p.ofs_unit;
- nsearched += count_bits(p.dirty_segmap,
- p.offset - p.ofs_unit,
- p.ofs_unit);
- } else {
- nsearched++;
- }
+ nsearched++;
#ifdef CONFIG_F2FS_CHECK_FS
/*
@@ -396,14 +724,35 @@ static int get_victim_by_default(struct f2fs_sb_info *sbi,
if (sec_usage_check(sbi, secno))
goto next;
+
/* Don't touch checkpointed data */
- if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED) &&
- get_ckpt_valid_blocks(sbi, segno) &&
- p.alloc_mode != SSR))
- goto next;
+ if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
+ if (p.alloc_mode == LFS) {
+ /*
+ * LFS is set to find source section during GC.
+ * The victim should have no checkpointed data.
+ */
+ if (get_ckpt_valid_blocks(sbi, segno, true))
+ goto next;
+ } else {
+ /*
+ * SSR | AT_SSR are set to find target segment
+ * for writes which can be full by checkpointed
+ * and newly written blocks.
+ */
+ if (!f2fs_segment_has_free_slot(sbi, segno))
+ goto next;
+ }
+ }
+
if (gc_type == BG_GC && test_bit(secno, dirty_i->victim_secmap))
goto next;
+ if (is_atgc) {
+ add_victim_entry(sbi, &p, segno);
+ goto next;
+ }
+
cost = get_gc_cost(sbi, segno, &p);
if (p.min_cost > cost) {
@@ -413,14 +762,28 @@ static int get_victim_by_default(struct f2fs_sb_info *sbi,
next:
if (nsearched >= p.max_search) {
if (!sm->last_victim[p.gc_mode] && segno <= last_victim)
- sm->last_victim[p.gc_mode] = last_victim + 1;
+ sm->last_victim[p.gc_mode] =
+ last_victim + p.ofs_unit;
else
- sm->last_victim[p.gc_mode] = segno + 1;
+ sm->last_victim[p.gc_mode] = segno + p.ofs_unit;
sm->last_victim[p.gc_mode] %=
(MAIN_SECS(sbi) * sbi->segs_per_sec);
break;
}
}
+
+ /* get victim for GC_AT/AT_SSR */
+ if (is_atgc) {
+ lookup_victim_by_age(sbi, &p);
+ release_victim_entry(sbi);
+ }
+
+ if (is_atgc && p.min_segno == NULL_SEGNO &&
+ sm->elapsed_time < p.age_threshold) {
+ p.age_threshold = 0;
+ goto retry;
+ }
+
if (p.min_segno != NULL_SEGNO) {
got_it:
*result = (p.min_segno / p.ofs_unit) * p.ofs_unit;
@@ -432,6 +795,7 @@ static int get_victim_by_default(struct f2fs_sb_info *sbi,
else
set_bit(secno, dirty_i->victim_secmap);
}
+ ret = 0;
}
out:
@@ -441,7 +805,7 @@ static int get_victim_by_default(struct f2fs_sb_info *sbi,
prefree_segments(sbi), free_segments(sbi));
mutex_unlock(&dirty_i->seglist_lock);
- return (p.min_segno == NULL_SEGNO) ? 0 : 1;
+ return ret;
}
static const struct victim_selection default_v_ops = {
@@ -512,6 +876,7 @@ static int gc_node_segment(struct f2fs_sb_info *sbi,
int phase = 0;
bool fggc = (gc_type == FG_GC);
int submitted = 0;
+ unsigned int usable_blks_in_seg = f2fs_usable_blks_in_seg(sbi, segno);
start_addr = START_BLOCK(sbi, segno);
@@ -521,7 +886,7 @@ static int gc_node_segment(struct f2fs_sb_info *sbi,
if (fggc && phase == 2)
atomic_inc(&sbi->wb_sync_req[NODE]);
- for (off = 0; off < sbi->blocks_per_seg; off++, entry++) {
+ for (off = 0; off < usable_blks_in_seg; off++, entry++) {
nid_t nid = le32_to_cpu(entry->nid);
struct page *node_page;
struct node_info ni;
@@ -633,8 +998,11 @@ static bool is_alive(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
set_sbi_flag(sbi, SBI_NEED_FSCK);
}
+ if (f2fs_check_nid_range(sbi, dni->ino))
+ return false;
+
*nofs = ofs_of_node(node_page);
- source_blkaddr = datablock_addr(NULL, node_page, ofs_in_node);
+ source_blkaddr = data_blkaddr(NULL, node_page, ofs_in_node);
f2fs_put_page(node_page, 1);
if (source_blkaddr != blkaddr) {
@@ -730,6 +1098,10 @@ static int ra_data_block(struct inode *inode, pgoff_t index)
goto put_encrypted_page;
f2fs_put_page(fio.encrypted_page, 0);
f2fs_put_page(page, 1);
+
+ f2fs_update_iostat(sbi, FS_DATA_READ_IO, F2FS_BLKSIZE);
+ f2fs_update_iostat(sbi, FS_GDATA_READ_IO, F2FS_BLKSIZE);
+
return 0;
put_encrypted_page:
f2fs_put_page(fio.encrypted_page, 1);
@@ -762,7 +1134,9 @@ static int move_data_block(struct inode *inode, block_t bidx,
struct page *page, *mpage;
block_t newaddr;
int err = 0;
- bool lfs_mode = test_opt(fio.sbi, LFS);
+ bool lfs_mode = f2fs_lfs_mode(fio.sbi);
+ int type = fio.sbi->am.atgc_enabled ?
+ CURSEG_ALL_DATA_ATGC : CURSEG_COLD_DATA;
/* do not read out */
page = f2fs_grab_cache_page(inode->i_mapping, bidx, false);
@@ -821,8 +1195,10 @@ static int move_data_block(struct inode *inode, block_t bidx,
mpage = f2fs_grab_cache_page(META_MAPPING(fio.sbi),
fio.old_blkaddr, false);
- if (!mpage)
+ if (!mpage) {
+ err = -ENOMEM;
goto up_out;
+ }
fio.encrypted_page = mpage;
@@ -833,6 +1209,10 @@ static int move_data_block(struct inode *inode, block_t bidx,
f2fs_put_page(mpage, 1);
goto up_out;
}
+
+ f2fs_update_iostat(fio.sbi, FS_DATA_READ_IO, F2FS_BLKSIZE);
+ f2fs_update_iostat(fio.sbi, FS_GDATA_READ_IO, F2FS_BLKSIZE);
+
lock_page(mpage);
if (unlikely(mpage->mapping != META_MAPPING(fio.sbi) ||
!PageUptodate(mpage))) {
@@ -843,7 +1223,7 @@ static int move_data_block(struct inode *inode, block_t bidx,
}
f2fs_allocate_data_block(fio.sbi, NULL, fio.old_blkaddr, &newaddr,
- &sum, CURSEG_COLD_DATA, NULL, false);
+ &sum, type, NULL);
fio.encrypted_page = f2fs_pagecache_get_page(META_MAPPING(fio.sbi),
newaddr, FGP_LOCK | FGP_CREAT, GFP_NOFS);
@@ -893,7 +1273,7 @@ static int move_data_block(struct inode *inode, block_t bidx,
recover_block:
if (err)
f2fs_do_replace_block(fio.sbi, &sum, newaddr, fio.old_blkaddr,
- true, true);
+ true, true, true);
up_out:
if (lfs_mode)
up_write(&fio.sbi->io_order_lock);
@@ -970,7 +1350,8 @@ static int move_data_page(struct inode *inode, block_t bidx, int gc_type,
if (err) {
clear_cold_data(page);
if (err == -ENOMEM) {
- congestion_wait(BLK_RW_ASYNC, HZ/50);
+ congestion_wait(BLK_RW_ASYNC,
+ DEFAULT_IO_TIMEOUT);
goto retry;
}
if (is_dirty)
@@ -990,7 +1371,8 @@ static int move_data_page(struct inode *inode, block_t bidx, int gc_type,
* the victim data block is ignored.
*/
static int gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
- struct gc_inode_list *gc_list, unsigned int segno, int gc_type)
+ struct gc_inode_list *gc_list, unsigned int segno, int gc_type,
+ bool force_migrate)
{
struct super_block *sb = sbi->sb;
struct f2fs_summary *entry;
@@ -998,13 +1380,14 @@ static int gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
int off;
int phase = 0;
int submitted = 0;
+ unsigned int usable_blks_in_seg = f2fs_usable_blks_in_seg(sbi, segno);
start_addr = START_BLOCK(sbi, segno);
next_step:
entry = sum;
- for (off = 0; off < sbi->blocks_per_seg; off++, entry++) {
+ for (off = 0; off < usable_blks_in_seg; off++, entry++) {
struct page *data_page;
struct inode *inode;
struct node_info dni; /* dnode info for the data */
@@ -1018,8 +1401,8 @@ static int gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
* race condition along with SSR block allocation.
*/
if ((gc_type == BG_GC && has_not_enough_free_secs(sbi, 0, 0)) ||
- get_valid_blocks(sbi, segno, false) ==
- sbi->blocks_per_seg)
+ (!force_migrate && get_valid_blocks(sbi, segno, true) ==
+ BLKS_PER_SEC(sbi)))
return submitted;
if (check_valid_map(sbi, segno, off) == 0)
@@ -1049,8 +1432,10 @@ static int gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
if (phase == 3) {
inode = f2fs_iget(sb, dni.ino);
- if (IS_ERR(inode) || is_bad_inode(inode))
+ if (IS_ERR(inode) || is_bad_inode(inode)) {
+ set_sbi_flag(sbi, SBI_NEED_FSCK);
continue;
+ }
if (!down_write_trylock(
&F2FS_I(inode)->i_gc_rwsem[WRITE])) {
@@ -1147,14 +1532,15 @@ static int __get_victim(struct f2fs_sb_info *sbi, unsigned int *victim,
down_write(&sit_i->sentry_lock);
ret = DIRTY_I(sbi)->v_ops->get_victim(sbi, victim, gc_type,
- NO_CHECK_TYPE, LFS);
+ NO_CHECK_TYPE, LFS, 0);
up_write(&sit_i->sentry_lock);
return ret;
}
static int do_garbage_collect(struct f2fs_sb_info *sbi,
unsigned int start_segno,
- struct gc_inode_list *gc_list, int gc_type)
+ struct gc_inode_list *gc_list, int gc_type,
+ bool force_migrate)
{
struct page *sum_page;
struct f2fs_summary_block *sum;
@@ -1169,6 +1555,17 @@ static int do_garbage_collect(struct f2fs_sb_info *sbi,
if (__is_large_section(sbi))
end_segno = rounddown(end_segno, sbi->segs_per_sec);
+ /*
+ * zone-capacity can be less than zone-size in zoned devices,
+ * resulting in less than expected usable segments in the zone,
+ * calculate the end segno in the zone which can be garbage collected
+ */
+ if (f2fs_sb_has_blkzoned(sbi))
+ end_segno -= sbi->segs_per_sec -
+ f2fs_usable_segs_in_sec(sbi, segno);
+
+ sanity_check_seg_type(sbi, get_seg_entry(sbi, segno)->type);
+
/* readahead multi ssa blocks those have contiguous address */
if (__is_large_section(sbi))
f2fs_ra_meta_pages(sbi, GET_SUM_BLOCK(sbi, segno),
@@ -1203,7 +1600,7 @@ static int do_garbage_collect(struct f2fs_sb_info *sbi,
if (get_valid_blocks(sbi, segno, false) == 0)
goto freed;
- if (__is_large_section(sbi) &&
+ if (gc_type == BG_GC && __is_large_section(sbi) &&
migrated >= sbi->migration_granularity)
goto skip;
if (!PageUptodate(sum_page) || unlikely(f2fs_cp_error(sbi)))
@@ -1230,15 +1627,16 @@ static int do_garbage_collect(struct f2fs_sb_info *sbi,
gc_type);
else
submitted += gc_data_segment(sbi, sum->entries, gc_list,
- segno, gc_type);
+ segno, gc_type,
+ force_migrate);
stat_inc_seg_count(sbi, type, gc_type);
+ migrated++;
freed:
if (gc_type == FG_GC &&
get_valid_blocks(sbi, segno, false) == 0)
seg_freed++;
- migrated++;
if (__is_large_section(sbi) && segno + 1 < end_segno)
sbi->next_victim_seg[gc_type] = segno + 1;
@@ -1258,7 +1656,7 @@ static int do_garbage_collect(struct f2fs_sb_info *sbi,
}
int f2fs_gc(struct f2fs_sb_info *sbi, bool sync,
- bool background, unsigned int segno)
+ bool background, bool force, unsigned int segno)
{
int gc_type = sync ? FG_GC : BG_GC;
int sec_freed = 0, seg_freed = 0, total_freed = 0;
@@ -1316,13 +1714,13 @@ int f2fs_gc(struct f2fs_sb_info *sbi, bool sync,
ret = -EINVAL;
goto stop;
}
- if (!__get_victim(sbi, &segno, gc_type)) {
- ret = -ENODATA;
+ ret = __get_victim(sbi, &segno, gc_type);
+ if (ret)
goto stop;
- }
- seg_freed = do_garbage_collect(sbi, segno, &gc_list, gc_type);
- if (gc_type == FG_GC && seg_freed == sbi->segs_per_sec)
+ seg_freed = do_garbage_collect(sbi, segno, &gc_list, gc_type, force);
+ if (gc_type == FG_GC &&
+ seg_freed == f2fs_usable_segs_in_sec(sbi, segno))
sec_freed++;
total_freed += seg_freed;
@@ -1370,7 +1768,7 @@ int f2fs_gc(struct f2fs_sb_info *sbi, bool sync,
reserved_segments(sbi),
prefree_segments(sbi));
- mutex_unlock(&sbi->gc_mutex);
+ up_write(&sbi->gc_lock);
put_gc_inode(&gc_list);
@@ -1379,6 +1777,38 @@ int f2fs_gc(struct f2fs_sb_info *sbi, bool sync,
return ret;
}
+int __init f2fs_create_garbage_collection_cache(void)
+{
+ victim_entry_slab = f2fs_kmem_cache_create("f2fs_victim_entry",
+ sizeof(struct victim_entry));
+ if (!victim_entry_slab)
+ return -ENOMEM;
+ return 0;
+}
+
+void f2fs_destroy_garbage_collection_cache(void)
+{
+ kmem_cache_destroy(victim_entry_slab);
+}
+
+static void init_atgc_management(struct f2fs_sb_info *sbi)
+{
+ struct atgc_management *am = &sbi->am;
+
+ if (test_opt(sbi, ATGC) &&
+ SIT_I(sbi)->elapsed_time >= DEF_GC_THREAD_AGE_THRESHOLD)
+ am->atgc_enabled = true;
+
+ am->root = RB_ROOT_CACHED;
+ INIT_LIST_HEAD(&am->victim_list);
+ am->victim_count = 0;
+
+ am->candidate_ratio = DEF_GC_THREAD_CANDIDATE_RATIO;
+ am->max_candidate_count = DEF_GC_THREAD_MAX_CANDIDATE_COUNT;
+ am->age_weight = DEF_GC_THREAD_AGE_WEIGHT;
+ am->age_threshold = DEF_GC_THREAD_AGE_THRESHOLD;
+}
+
void f2fs_build_gc_manager(struct f2fs_sb_info *sbi)
{
DIRTY_I(sbi)->v_ops = &default_v_ops;
@@ -1389,18 +1819,37 @@ void f2fs_build_gc_manager(struct f2fs_sb_info *sbi)
if (f2fs_is_multi_device(sbi) && !__is_large_section(sbi))
SIT_I(sbi)->last_victim[ALLOC_NEXT] =
GET_SEGNO(sbi, FDEV(0).end_blk) + 1;
+
+ init_atgc_management(sbi);
}
-static int free_segment_range(struct f2fs_sb_info *sbi, unsigned int start,
- unsigned int end)
+static int free_segment_range(struct f2fs_sb_info *sbi,
+ unsigned int secs, bool gc_only)
{
- int type;
- unsigned int segno, next_inuse;
+ unsigned int segno, next_inuse, start, end;
+ struct cp_control cpc = { CP_RESIZE, 0, 0, 0 };
+ int gc_mode, gc_type;
int err = 0;
+ int type;
+
+ /* Force block allocation for GC */
+ MAIN_SECS(sbi) -= secs;
+ start = MAIN_SECS(sbi) * sbi->segs_per_sec;
+ end = MAIN_SEGS(sbi) - 1;
+
+ mutex_lock(&DIRTY_I(sbi)->seglist_lock);
+ for (gc_mode = 0; gc_mode < MAX_GC_POLICY; gc_mode++)
+ if (SIT_I(sbi)->last_victim[gc_mode] >= start)
+ SIT_I(sbi)->last_victim[gc_mode] = 0;
+
+ for (gc_type = BG_GC; gc_type <= FG_GC; gc_type++)
+ if (sbi->next_victim_seg[gc_type] >= start)
+ sbi->next_victim_seg[gc_type] = NULL_SEGNO;
+ mutex_unlock(&DIRTY_I(sbi)->seglist_lock);
/* Move out cursegs from the target range */
- for (type = CURSEG_HOT_DATA; type < NR_CURSEG_TYPE; type++)
- allocate_segment_for_resize(sbi, type, start, end);
+ for (type = CURSEG_HOT_DATA; type < NR_CURSEG_PERSIST_TYPE; type++)
+ f2fs_allocate_segment_for_resize(sbi, type, start, end);
/* do GC to move out valid blocks in the range */
for (segno = start; segno <= end; segno += sbi->segs_per_sec) {
@@ -1409,18 +1858,24 @@ static int free_segment_range(struct f2fs_sb_info *sbi, unsigned int start,
.iroot = RADIX_TREE_INIT(gc_list.iroot, GFP_NOFS),
};
- mutex_lock(&sbi->gc_mutex);
- do_garbage_collect(sbi, segno, &gc_list, FG_GC);
- mutex_unlock(&sbi->gc_mutex);
+ do_garbage_collect(sbi, segno, &gc_list, FG_GC, true);
put_gc_inode(&gc_list);
- if (get_valid_blocks(sbi, segno, true))
- return -EAGAIN;
+ if (!gc_only && get_valid_blocks(sbi, segno, true)) {
+ err = -EAGAIN;
+ goto out;
+ }
+ if (fatal_signal_pending(current)) {
+ err = -ERESTARTSYS;
+ goto out;
+ }
}
+ if (gc_only)
+ goto out;
- err = f2fs_sync_fs(sbi->sb, 1);
+ err = f2fs_write_checkpoint(sbi, &cpc);
if (err)
- return err;
+ goto out;
next_inuse = find_next_inuse(FREE_I(sbi), end + 1, start);
if (next_inuse <= end) {
@@ -1428,44 +1883,77 @@ static int free_segment_range(struct f2fs_sb_info *sbi, unsigned int start,
next_inuse);
f2fs_bug_on(sbi, 1);
}
+out:
+ MAIN_SECS(sbi) += secs;
return err;
}
static void update_sb_metadata(struct f2fs_sb_info *sbi, int secs)
{
struct f2fs_super_block *raw_sb = F2FS_RAW_SUPER(sbi);
- int section_count = le32_to_cpu(raw_sb->section_count);
- int segment_count = le32_to_cpu(raw_sb->segment_count);
- int segment_count_main = le32_to_cpu(raw_sb->segment_count_main);
- long long block_count = le64_to_cpu(raw_sb->block_count);
+ int section_count;
+ int segment_count;
+ int segment_count_main;
+ long long block_count;
int segs = secs * sbi->segs_per_sec;
+ down_write(&sbi->sb_lock);
+
+ section_count = le32_to_cpu(raw_sb->section_count);
+ segment_count = le32_to_cpu(raw_sb->segment_count);
+ segment_count_main = le32_to_cpu(raw_sb->segment_count_main);
+ block_count = le64_to_cpu(raw_sb->block_count);
+
raw_sb->section_count = cpu_to_le32(section_count + secs);
raw_sb->segment_count = cpu_to_le32(segment_count + segs);
raw_sb->segment_count_main = cpu_to_le32(segment_count_main + segs);
raw_sb->block_count = cpu_to_le64(block_count +
(long long)segs * sbi->blocks_per_seg);
+ if (f2fs_is_multi_device(sbi)) {
+ int last_dev = sbi->s_ndevs - 1;
+ int dev_segs =
+ le32_to_cpu(raw_sb->devs[last_dev].total_segments);
+
+ raw_sb->devs[last_dev].total_segments =
+ cpu_to_le32(dev_segs + segs);
+ }
+
+ up_write(&sbi->sb_lock);
}
static void update_fs_metadata(struct f2fs_sb_info *sbi, int secs)
{
int segs = secs * sbi->segs_per_sec;
+ long long blks = (long long)segs * sbi->blocks_per_seg;
long long user_block_count =
le64_to_cpu(F2FS_CKPT(sbi)->user_block_count);
SM_I(sbi)->segment_count = (int)SM_I(sbi)->segment_count + segs;
MAIN_SEGS(sbi) = (int)MAIN_SEGS(sbi) + segs;
+ MAIN_SECS(sbi) += secs;
FREE_I(sbi)->free_sections = (int)FREE_I(sbi)->free_sections + secs;
FREE_I(sbi)->free_segments = (int)FREE_I(sbi)->free_segments + segs;
- F2FS_CKPT(sbi)->user_block_count = cpu_to_le64(user_block_count +
- (long long)segs * sbi->blocks_per_seg);
+ F2FS_CKPT(sbi)->user_block_count = cpu_to_le64(user_block_count + blks);
+
+ if (f2fs_is_multi_device(sbi)) {
+ int last_dev = sbi->s_ndevs - 1;
+
+ FDEV(last_dev).total_segments =
+ (int)FDEV(last_dev).total_segments + segs;
+ FDEV(last_dev).end_blk =
+ (long long)FDEV(last_dev).end_blk + blks;
+#ifdef CONFIG_BLK_DEV_ZONED
+ FDEV(last_dev).nr_blkz = (int)FDEV(last_dev).nr_blkz +
+ (int)(blks >> sbi->log_blocks_per_blkz);
+#endif
+ }
}
int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count)
{
__u64 old_block_count, shrunk_blocks;
+ struct cp_control cpc = { CP_RESIZE, 0, 0, 0 };
unsigned int secs;
- int gc_mode, gc_type;
int err = 0;
__u32 rem;
@@ -1473,6 +1961,15 @@ int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count)
if (block_count > old_block_count)
return -EINVAL;
+ if (f2fs_is_multi_device(sbi)) {
+ int last_dev = sbi->s_ndevs - 1;
+ __u64 last_segs = FDEV(last_dev).total_segments;
+
+ if (block_count + last_segs * sbi->blocks_per_seg <=
+ old_block_count)
+ return -EINVAL;
+ }
+
/* new fs size should align to section size */
div_u64_rem(block_count, BLKS_PER_SEC(sbi), &rem);
if (rem)
@@ -1491,10 +1988,40 @@ int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count)
return -EINVAL;
}
- freeze_bdev(sbi->sb->s_bdev);
-
shrunk_blocks = old_block_count - block_count;
secs = div_u64(shrunk_blocks, BLKS_PER_SEC(sbi));
+
+ /* stop other GC */
+ if (!down_write_trylock(&sbi->gc_lock))
+ return -EAGAIN;
+
+ /* stop CP to protect MAIN_SEC in free_segment_range */
+ f2fs_lock_op(sbi);
+
+ spin_lock(&sbi->stat_lock);
+ if (shrunk_blocks + valid_user_blocks(sbi) +
+ sbi->current_reserved_blocks + sbi->unusable_block_count +
+ F2FS_OPTION(sbi).root_reserved_blocks > sbi->user_block_count)
+ err = -ENOSPC;
+ spin_unlock(&sbi->stat_lock);
+
+ if (err)
+ goto out_unlock;
+
+ err = free_segment_range(sbi, secs, true);
+
+out_unlock:
+ f2fs_unlock_op(sbi);
+ up_write(&sbi->gc_lock);
+ if (err)
+ return err;
+
+ set_sbi_flag(sbi, SBI_IS_RESIZEFS);
+
+ freeze_super(sbi->sb);
+ down_write(&sbi->gc_lock);
+ mutex_lock(&sbi->cp_mutex);
+
spin_lock(&sbi->stat_lock);
if (shrunk_blocks + valid_user_blocks(sbi) +
sbi->current_reserved_blocks + sbi->unusable_block_count +
@@ -1503,69 +2030,44 @@ int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count)
else
sbi->user_block_count -= shrunk_blocks;
spin_unlock(&sbi->stat_lock);
- if (err) {
- thaw_bdev(sbi->sb->s_bdev, sbi->sb);
- return err;
- }
-
- mutex_lock(&sbi->resize_mutex);
- set_sbi_flag(sbi, SBI_IS_RESIZEFS);
-
- mutex_lock(&DIRTY_I(sbi)->seglist_lock);
-
- MAIN_SECS(sbi) -= secs;
-
- for (gc_mode = 0; gc_mode < MAX_GC_POLICY; gc_mode++)
- if (SIT_I(sbi)->last_victim[gc_mode] >=
- MAIN_SECS(sbi) * sbi->segs_per_sec)
- SIT_I(sbi)->last_victim[gc_mode] = 0;
-
- for (gc_type = BG_GC; gc_type <= FG_GC; gc_type++)
- if (sbi->next_victim_seg[gc_type] >=
- MAIN_SECS(sbi) * sbi->segs_per_sec)
- sbi->next_victim_seg[gc_type] = NULL_SEGNO;
-
- mutex_unlock(&DIRTY_I(sbi)->seglist_lock);
-
- err = free_segment_range(sbi, MAIN_SECS(sbi) * sbi->segs_per_sec,
- MAIN_SEGS(sbi) - 1);
if (err)
- goto out;
+ goto out_err;
+
+ err = free_segment_range(sbi, secs, false);
+ if (err)
+ goto recover_out;
update_sb_metadata(sbi, -secs);
err = f2fs_commit_super(sbi, false);
if (err) {
update_sb_metadata(sbi, secs);
- goto out;
+ goto recover_out;
}
- mutex_lock(&sbi->cp_mutex);
update_fs_metadata(sbi, -secs);
clear_sbi_flag(sbi, SBI_IS_RESIZEFS);
set_sbi_flag(sbi, SBI_IS_DIRTY);
- mutex_unlock(&sbi->cp_mutex);
- err = f2fs_sync_fs(sbi->sb, 1);
+ err = f2fs_write_checkpoint(sbi, &cpc);
if (err) {
- mutex_lock(&sbi->cp_mutex);
update_fs_metadata(sbi, secs);
- mutex_unlock(&sbi->cp_mutex);
update_sb_metadata(sbi, secs);
f2fs_commit_super(sbi, false);
}
-out:
+recover_out:
if (err) {
set_sbi_flag(sbi, SBI_NEED_FSCK);
f2fs_err(sbi, "resize_fs failed, should run fsck to repair!");
- MAIN_SECS(sbi) += secs;
spin_lock(&sbi->stat_lock);
sbi->user_block_count += shrunk_blocks;
spin_unlock(&sbi->stat_lock);
}
+out_err:
+ mutex_unlock(&sbi->cp_mutex);
+ up_write(&sbi->gc_lock);
+ thaw_super(sbi->sb);
clear_sbi_flag(sbi, SBI_IS_RESIZEFS);
- mutex_unlock(&sbi->resize_mutex);
- thaw_bdev(sbi->sb->s_bdev, sbi->sb);
return err;
}