f2fs: add a sysfs entry to reclaim POSIX_FADV_NOREUSE pages

1. fadvise(fd1, POSIX_FADV_NOREUSE, {0,3});
2. fadvise(fd2, POSIX_FADV_NOREUSE, {1,2});
3. fadvise(fd3, POSIX_FADV_NOREUSE, {3,1});
4. echo 1024 > /sys/fs/f2fs/tuning/reclaim_caches_kb

This gives a way to reclaim file-backed pages by iterating all f2fs mounts until
reclaiming 1MB page cache ranges, registered by #1, #2, and #3.

5. cat /sys/fs/f2fs/tuning/reclaim_caches_kb
-> gives total number of registered file ranges.

Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>

diff --git a/Documentation/ABI/testing/sysfs-fs-f2fs b/Documentation/ABI/testing/sysfs-fs-f2fs
index 3e1630c70d8ae70456de35a627afde1211c8882c..81deae2af84d2a6b7b2999dfb3c86c9a246f3131 100644 (file)
--- a/Documentation/ABI/testing/sysfs-fs-f2fs
+++ b/Documentation/ABI/testing/sysfs-fs-f2fs
@@ -828,3 +828,10 @@ Date:              November 2024
 Contact:       "Chao Yu" <chao@kernel.org>
 Description:   It controls max read extent count for per-inode, the value of threshold
                is 10240 by default.
+
+What:          /sys/fs/f2fs/tuning/reclaim_caches_kb
+Date:          February 2025
+Contact:       "Jaegeuk Kim" <jaegeuk@kernel.org>
+Description:   It reclaims the given KBs of file-backed pages registered by
+               ioctl(F2FS_IOC_DONATE_RANGE).
+               For example, writing N tries to drop N KBs spaces in LRU.

diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
index 3abcb84a0d47493fb2d6d1bacfa9cf4f958cdda9..05879c6dc4d6c77aea9f831c11745552626ad065 100644 (file)
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -4243,6 +4243,8 @@ unsigned long f2fs_shrink_count(struct shrinker *shrink,
                        struct shrink_control *sc);
 unsigned long f2fs_shrink_scan(struct shrinker *shrink,
                        struct shrink_control *sc);
+unsigned int f2fs_donate_files(void);
+void f2fs_reclaim_caches(unsigned int reclaim_caches_kb);
 void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
 

diff --git a/fs/f2fs/shrinker.c b/fs/f2fs/shrinker.c
index fc1bbef418ce930cedf0e823d29caed1f168768e..9c8d3aee89afa022fbfa969858b1358e3c85ac05 100644 (file)
--- a/fs/f2fs/shrinker.c
+++ b/fs/f2fs/shrinker.c
@@ -130,6 +130,96 @@ unsigned long f2fs_shrink_scan(struct shrinker *shrink,
        return freed;
 }
 
+unsigned int f2fs_donate_files(void)
+{
+       struct f2fs_sb_info *sbi;
+       struct list_head *p;
+       unsigned int donate_files = 0;
+
+       spin_lock(&f2fs_list_lock);
+       p = f2fs_list.next;
+       while (p != &f2fs_list) {
+               sbi = list_entry(p, struct f2fs_sb_info, s_list);
+
+               /* stop f2fs_put_super */
+               if (!mutex_trylock(&sbi->umount_mutex)) {
+                       p = p->next;
+                       continue;
+               }
+               spin_unlock(&f2fs_list_lock);
+
+               donate_files += sbi->donate_files;
+
+               spin_lock(&f2fs_list_lock);
+               p = p->next;
+               mutex_unlock(&sbi->umount_mutex);
+       }
+       spin_unlock(&f2fs_list_lock);
+
+       return donate_files;
+}
+
+static unsigned int do_reclaim_caches(struct f2fs_sb_info *sbi,
+                               unsigned int reclaim_caches_kb)
+{
+       struct inode *inode;
+       struct f2fs_inode_info *fi;
+       unsigned int nfiles = sbi->donate_files;
+       pgoff_t npages = reclaim_caches_kb >> (PAGE_SHIFT - 10);
+
+       while (npages && nfiles--) {
+               pgoff_t len;
+
+               spin_lock(&sbi->inode_lock[DONATE_INODE]);
+               if (list_empty(&sbi->inode_list[DONATE_INODE])) {
+                       spin_unlock(&sbi->inode_lock[DONATE_INODE]);
+                       break;
+               }
+               fi = list_first_entry(&sbi->inode_list[DONATE_INODE],
+                                       struct f2fs_inode_info, gdonate_list);
+               list_move_tail(&fi->gdonate_list, &sbi->inode_list[DONATE_INODE]);
+               inode = igrab(&fi->vfs_inode);
+               spin_unlock(&sbi->inode_lock[DONATE_INODE]);
+
+               if (!inode)
+                       continue;
+
+               len = fi->donate_end - fi->donate_start + 1;
+               npages = npages < len ? 0 : npages - len;
+               invalidate_inode_pages2_range(inode->i_mapping,
+                                       fi->donate_start, fi->donate_end);
+               iput(inode);
+               cond_resched();
+       }
+       return npages << (PAGE_SHIFT - 10);
+}
+
+void f2fs_reclaim_caches(unsigned int reclaim_caches_kb)
+{
+       struct f2fs_sb_info *sbi;
+       struct list_head *p;
+
+       spin_lock(&f2fs_list_lock);
+       p = f2fs_list.next;
+       while (p != &f2fs_list && reclaim_caches_kb) {
+               sbi = list_entry(p, struct f2fs_sb_info, s_list);
+
+               /* stop f2fs_put_super */
+               if (!mutex_trylock(&sbi->umount_mutex)) {
+                       p = p->next;
+                       continue;
+               }
+               spin_unlock(&f2fs_list_lock);
+
+               reclaim_caches_kb = do_reclaim_caches(sbi, reclaim_caches_kb);
+
+               spin_lock(&f2fs_list_lock);
+               p = p->next;
+               mutex_unlock(&sbi->umount_mutex);
+       }
+       spin_unlock(&f2fs_list_lock);
+}
+
 void f2fs_join_shrinker(struct f2fs_sb_info *sbi)
 {
        spin_lock(&f2fs_list_lock);

diff --git a/fs/f2fs/sysfs.c b/fs/f2fs/sysfs.c
index b419555e1ea7f7a0dbd5270ed3f00015a2c743e4..b27336acf519e9a944bbd91cfdcb96235091f662 100644 (file)
--- a/fs/f2fs/sysfs.c
+++ b/fs/f2fs/sysfs.c
@@ -916,6 +916,39 @@ static struct f2fs_base_attr f2fs_base_attr_##_name = {            \
        .show   = f2fs_feature_show,                            \
 }
 
+static ssize_t f2fs_tune_show(struct f2fs_base_attr *a, char *buf)
+{
+       unsigned int res = 0;
+
+       if (!strcmp(a->attr.name, "reclaim_caches_kb"))
+               res = f2fs_donate_files();
+
+       return sysfs_emit(buf, "%u\n", res);
+}
+
+static ssize_t f2fs_tune_store(struct f2fs_base_attr *a,
+                       const char *buf, size_t count)
+{
+       unsigned long t;
+       int ret;
+
+       ret = kstrtoul(skip_spaces(buf), 0, &t);
+       if (ret)
+               return ret;
+
+       if (!strcmp(a->attr.name, "reclaim_caches_kb"))
+               f2fs_reclaim_caches(t);
+
+       return count;
+}
+
+#define F2FS_TUNE_RW_ATTR(_name)                               \
+static struct f2fs_base_attr f2fs_base_attr_##_name = {                \
+       .attr = {.name = __stringify(_name), .mode = 0644 },    \
+       .show   = f2fs_tune_show,                               \
+       .store  = f2fs_tune_store,                              \
+}
+
 static ssize_t f2fs_sb_feature_show(struct f2fs_attr *a,
                struct f2fs_sb_info *sbi, char *buf)
 {
@@ -1368,6 +1401,14 @@ static struct attribute *f2fs_sb_feat_attrs[] = {
 };
 ATTRIBUTE_GROUPS(f2fs_sb_feat);
 
+F2FS_TUNE_RW_ATTR(reclaim_caches_kb);
+
+static struct attribute *f2fs_tune_attrs[] = {
+       BASE_ATTR_LIST(reclaim_caches_kb),
+       NULL,
+};
+ATTRIBUTE_GROUPS(f2fs_tune);
+
 static const struct sysfs_ops f2fs_attr_ops = {
        .show   = f2fs_attr_show,
        .store  = f2fs_attr_store,
@@ -1401,6 +1442,20 @@ static struct kobject f2fs_feat = {
        .kset   = &f2fs_kset,
 };
 
+static const struct sysfs_ops f2fs_tune_attr_ops = {
+       .show   = f2fs_base_attr_show,
+       .store  = f2fs_base_attr_store,
+};
+
+static const struct kobj_type f2fs_tune_ktype = {
+       .default_groups = f2fs_tune_groups,
+       .sysfs_ops      = &f2fs_tune_attr_ops,
+};
+
+static struct kobject f2fs_tune = {
+       .kset   = &f2fs_kset,
+};
+
 static ssize_t f2fs_stat_attr_show(struct kobject *kobj,
                                struct attribute *attr, char *buf)
 {
@@ -1637,6 +1692,11 @@ int __init f2fs_init_sysfs(void)
        if (ret)
                goto put_kobject;
 
+       ret = kobject_init_and_add(&f2fs_tune, &f2fs_tune_ktype,
+                                  NULL, "tuning");
+       if (ret)
+               goto put_kobject;
+
        f2fs_proc_root = proc_mkdir("fs/f2fs", NULL);
        if (!f2fs_proc_root) {
                ret = -ENOMEM;
@@ -1644,7 +1704,9 @@ int __init f2fs_init_sysfs(void)
        }
 
        return 0;
+
 put_kobject:
+       kobject_put(&f2fs_tune);
        kobject_put(&f2fs_feat);
        kset_unregister(&f2fs_kset);
        return ret;
@@ -1652,6 +1714,7 @@ put_kobject:
 
 void f2fs_exit_sysfs(void)
 {
+       kobject_put(&f2fs_tune);
        kobject_put(&f2fs_feat);
        kset_unregister(&f2fs_kset);
        remove_proc_entry("fs/f2fs", NULL);