| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <4df93910-d942-2410-f7cd-cc28f7a00a86@gmail.com>
Date: Wed, 26 May 2021 16:43:51 +0800
From: Wang Jianchao <jianchao.wan9@...il.com>
To: Theodore Ts'o <tytso@....edu>,
Andreas Dilger <adilger.kernel@...ger.ca>
Cc: linux-ext4@...r.kernel.org, linux-kernel@...r.kernel.org,
lishujin@...ishou.com
Subject: [PATCH V2 6/7] ext4: use bb_free_root to get the free data entry
This is also preparing for following async background discard.
In this patch, the s_freed_data_list is removed and we iterate
all of the group's free_data_root rb tree to get the entry.
After this, we needn't operate it when insert and merge free
data entry any more.
Signed-off-by: Wang Jianchao <wangjianchao@...ishou.com>
---
fs/ext4/balloc.c | 2 +-
fs/ext4/ext4.h | 4 +--
fs/ext4/mballoc.c | 104 +++++++++++++++++++++++++-----------------------------
3 files changed, 50 insertions(+), 60 deletions(-)
diff --git a/fs/ext4/balloc.c b/fs/ext4/balloc.c
index 74a5172..8053a5f 100644
--- a/fs/ext4/balloc.c
+++ b/fs/ext4/balloc.c
@@ -652,7 +652,7 @@ int ext4_should_retry_alloc(struct super_block *sb, int *retries)
* possible we just missed a transaction commit that did so
*/
smp_mb();
- if (sbi->s_mb_free_pending == 0)
+ if (!atomic_read(&sbi->s_mb_free_pending))
return ext4_has_free_clusters(sbi, 1, 0);
/*
diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index 826a56e3..5c5c8e4 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -1525,9 +1525,7 @@ struct ext4_sb_info {
unsigned short *s_mb_offsets;
unsigned int *s_mb_maxs;
unsigned int s_group_info_size;
- unsigned int s_mb_free_pending;
- struct list_head s_freed_data_list; /* List of blocks to be freed
- after commit completed */
+ atomic_t s_mb_free_pending;
/* tunables */
unsigned long s_stripe;
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index c2bf40a..15715e7 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -354,8 +354,7 @@ static inline struct ext4_free_data *efd_entry(struct rb_node *n)
{
return rb_entry_safe(n, struct ext4_free_data, efd_node);
}
-static int ext4_insert_free_data(struct ext4_sb_info *sbi,
- struct rb_root *root, struct ext4_free_data *nfd);
+static int ext4_insert_free_data(struct rb_root *root, struct ext4_free_data *nfd);
/*
* The algorithm using this percpu seq counter goes below:
@@ -2857,8 +2856,7 @@ int ext4_mb_init(struct super_block *sb)
spin_lock_init(&sbi->s_md_lock);
spin_lock_init(&sbi->s_bal_lock);
- sbi->s_mb_free_pending = 0;
- INIT_LIST_HEAD(&sbi->s_freed_data_list);
+ atomic_set(&sbi->s_mb_free_pending, 0);
sbi->s_mb_max_to_scan = MB_DEFAULT_MAX_TO_SCAN;
sbi->s_mb_min_to_scan = MB_DEFAULT_MIN_TO_SCAN;
@@ -3040,9 +3038,7 @@ static void ext4_free_data_in_buddy(struct super_block *sb,
/* we expect to find existing buddy because it's pinned */
BUG_ON(err != 0);
- spin_lock(&EXT4_SB(sb)->s_md_lock);
- EXT4_SB(sb)->s_mb_free_pending -= entry->efd_count;
- spin_unlock(&EXT4_SB(sb)->s_md_lock);
+ atomic_sub(entry->efd_count, &EXT4_SB(sb)->s_mb_free_pending);
db = e4b.bd_info;
/* there are blocks to put in buddy to make them really free */
@@ -3084,37 +3080,41 @@ static void ext4_free_data_in_buddy(struct super_block *sb,
void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid)
{
struct ext4_sb_info *sbi = EXT4_SB(sb);
- struct ext4_free_data *entry, *tmp;
+ ext4_group_t ngroups = ext4_get_groups_count(sb);
+ struct ext4_free_data *fd, *nfd;
+ struct ext4_group_info *grp;
struct bio *discard_bio = NULL;
struct list_head freed_data_list;
- struct list_head *cut_pos = NULL;
- int err;
+ int err, i;
- INIT_LIST_HEAD(&freed_data_list);
+ if (!atomic_read(&sbi->s_mb_free_pending))
+ return;
- spin_lock(&sbi->s_md_lock);
- list_for_each_entry(entry, &sbi->s_freed_data_list, efd_list) {
- if (entry->efd_tid != commit_tid)
- break;
- cut_pos = &entry->efd_list;
+ INIT_LIST_HEAD(&freed_data_list);
+ for (i = 0; i < ngroups; i++) {
+ grp = ext4_get_group_info(sb, i);
+ ext4_lock_group(sb, i);
+ rbtree_postorder_for_each_entry_safe(fd, nfd,
+ &grp->bb_free_root, efd_node) {
+ if (fd->efd_tid != commit_tid)
+ continue;
+ INIT_LIST_HEAD(&fd->efd_list);
+ list_add_tail(&fd->efd_list, &freed_data_list);
+ }
+ ext4_unlock_group(sb, i);
}
- if (cut_pos)
- list_cut_position(&freed_data_list, &sbi->s_freed_data_list,
- cut_pos);
- spin_unlock(&sbi->s_md_lock);
if (test_opt(sb, DISCARD)) {
- list_for_each_entry(entry, &freed_data_list, efd_list) {
- err = ext4_issue_discard(sb, entry->efd_group,
- entry->efd_start_cluster,
- entry->efd_count,
+ list_for_each_entry(fd, &freed_data_list, efd_list) {
+ err = ext4_issue_discard(sb, fd->efd_group,
+ fd->efd_start_cluster,
+ fd->efd_count,
&discard_bio);
if (err && err != -EOPNOTSUPP) {
ext4_msg(sb, KERN_WARNING, "discard request in"
" group:%d block:%d count:%d failed"
- " with %d", entry->efd_group,
- entry->efd_start_cluster,
- entry->efd_count, err);
+ " with %d", fd->efd_group,
+ fd->efd_start_cluster, fd->efd_count, err);
} else if (err == -EOPNOTSUPP)
break;
}
@@ -3125,8 +3125,8 @@ void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid)
}
}
- list_for_each_entry_safe(entry, tmp, &freed_data_list, efd_list)
- ext4_free_data_in_buddy(sb, entry);
+ list_for_each_entry_safe(fd, nfd, &freed_data_list, efd_list)
+ ext4_free_data_in_buddy(sb, fd);
}
int __init ext4_init_mballoc(void)
@@ -5051,32 +5051,27 @@ ext4_fsblk_t ext4_mb_new_blocks(handle_t *handle,
* are contiguous, AND the extents were freed by the same transaction,
* AND the blocks are associated with the same group.
*/
-static void ext4_try_merge_freed_extent(struct ext4_sb_info *sbi,
- struct ext4_free_data *entry,
- struct ext4_free_data *new_entry,
- struct rb_root *entry_rb_root)
+static void ext4_try_merge_freed_extent(struct rb_root *root,
+ struct ext4_free_data *fd, struct ext4_free_data *nfd)
{
- if ((entry->efd_tid != new_entry->efd_tid) ||
- (entry->efd_group != new_entry->efd_group))
+ if ((fd->efd_tid != nfd->efd_tid) ||
+ (fd->efd_group != nfd->efd_group))
return;
- if (entry->efd_start_cluster + entry->efd_count ==
- new_entry->efd_start_cluster) {
- new_entry->efd_start_cluster = entry->efd_start_cluster;
- new_entry->efd_count += entry->efd_count;
- } else if (new_entry->efd_start_cluster + new_entry->efd_count ==
- entry->efd_start_cluster) {
- new_entry->efd_count += entry->efd_count;
+ if (fd->efd_start_cluster + fd->efd_count ==
+ nfd->efd_start_cluster) {
+ nfd->efd_start_cluster = fd->efd_start_cluster;
+ nfd->efd_count += fd->efd_count;
+ } else if (nfd->efd_start_cluster + nfd->efd_count ==
+ fd->efd_start_cluster) {
+ nfd->efd_count += fd->efd_count;
} else
return;
- spin_lock(&sbi->s_md_lock);
- list_del(&entry->efd_list);
- spin_unlock(&sbi->s_md_lock);
- rb_erase(&entry->efd_node, entry_rb_root);
- kmem_cache_free(ext4_free_data_cachep, entry);
+ rb_erase(&fd->efd_node, root);
+ kmem_cache_free(ext4_free_data_cachep, fd);
}
-static int ext4_insert_free_data(struct ext4_sb_info *sbi,
- struct rb_root *root, struct ext4_free_data *nfd)
+static int ext4_insert_free_data(struct rb_root *root,
+ struct ext4_free_data *nfd)
{
struct rb_node **n = &root->rb_node;
struct rb_node *p = NULL;
@@ -5100,11 +5095,11 @@ static int ext4_insert_free_data(struct ext4_sb_info *sbi,
/* Now try to see the extent can be merged to left and right */
fd = efd_entry(rb_prev(&nfd->efd_node));
if (fd)
- ext4_try_merge_freed_extent(sbi, fd, nfd, root);
+ ext4_try_merge_freed_extent(root, fd, nfd);
fd = efd_entry(rb_next(&nfd->efd_node));
if (fd)
- ext4_try_merge_freed_extent(sbi, fd, nfd, root);
+ ext4_try_merge_freed_extent(root, fd, nfd);
return 0;
}
@@ -5122,7 +5117,7 @@ static int ext4_insert_free_data(struct ext4_sb_info *sbi,
BUG_ON(e4b->bd_bitmap_page == NULL);
BUG_ON(e4b->bd_buddy_page == NULL);
- if (ext4_insert_free_data(sbi, &db->bb_free_root, nfd)) {
+ if (ext4_insert_free_data(&db->bb_free_root, nfd)) {
ext4_grp_locked_error(sb, e4b->bd_group, 0,
ext4_group_first_block_no(sb, e4b->bd_group) +
EXT4_C2B(sbi, nfd->efd_start_cluster),
@@ -5140,10 +5135,7 @@ static int ext4_insert_free_data(struct ext4_sb_info *sbi,
get_page(e4b->bd_bitmap_page);
}
- spin_lock(&sbi->s_md_lock);
- list_add_tail(&nfd->efd_list, &sbi->s_freed_data_list);
- sbi->s_mb_free_pending += nfd->efd_count;
- spin_unlock(&sbi->s_md_lock);
+ atomic_add(nfd->efd_count, &sbi->s_mb_free_pending);
return 0;
}
--
1.8.3.1
Powered by blists - more mailing lists