f2fs: ioctl for removing a range from F2FS

This ioctl shrinks a given length (aligned to sections) from end of the
main area. Any cursegs and valid blocks will be moved out before
invalidating the range.

This feature can be used for adjusting partition sizes online.

History of the patch:

Sahitya Tummala:
 - Add this ioctl for f2fs_compat_ioctl() as well.
 - Fix debugfs status to reflect the online resize changes.
 - Fix potential race between online resize path and allocate new data
   block path or gc path.

Others:
 - Rename some identifiers.
 - Add some error handling branches.
 - Clear sbi->next_victim_seg[BG_GC/FG_GC] in shrinking range.
 - Implement this interface as ext4's, and change the parameter from shrunk
bytes to new block count of F2FS.
 - During resizing, force to empty sit_journal and forbid adding new
   entries to it, in order to avoid invalid segno in journal after resize.
 - Reduce sbi->user_block_count before resize starts.
 - Commit the updated superblock first, and then update in-memory metadata
   only when the former succeeds.
 - Target block count must align to sections.
 - Write checkpoint before and after committing the new superblock, w/o
CP_FSCK_FLAG respectively, so that the FS can be fixed by fsck even if
resize fails after the new superblock is committed.
 - In free_segment_range(), reduce granularity of gc_mutex.
 - Add protection on curseg migration.
 - Add freeze_bdev() and thaw_bdev() for resize fs.
 - Remove CUR_MAIN_SECS and use MAIN_SECS directly for allocation.
 - Recover super_block and FS metadata when resize fails.
 - No need to clear CP_FSCK_FLAG in update_ckpt_flags().
 - Clean up the sb and fs metadata update functions for resize_fs.

Geert Uytterhoeven:
 - Use div_u64*() for 64-bit divisions

Arnd Bergmann:
 - Not all architectures support get_user() with a 64-bit argument:
    ERROR: "__get_user_bad" [fs/f2fs/f2fs.ko] undefined!
    Use copy_from_user() here, this will always work.

Signed-off-by: Qiuyang Sun <sunqiuyang@huawei.com>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Sahitya Tummala <stummala@codeaurora.org>
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>

1 files changed, 180 insertions, 2 deletions

diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c
index 1e029da26053..e19b49b02d1b 100644
--- a/fs/f2fs/gc.c
+++ b/fs/f2fs/gc.c
@@ -311,10 +311,11 @@ static int get_victim_by_default(struct f2fs_sb_info *sbi,
 	struct sit_info *sm = SIT_I(sbi);
 	struct victim_sel_policy p;
 	unsigned int secno, last_victim;
-	unsigned int last_segment = MAIN_SEGS(sbi);
+	unsigned int last_segment;
 	unsigned int nsearched = 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);
@@ -405,7 +406,8 @@ next:
 				sm->last_victim[p.gc_mode] = last_victim + 1;
 			else
 				sm->last_victim[p.gc_mode] = segno + 1;
-			sm->last_victim[p.gc_mode] %= MAIN_SEGS(sbi);
+			sm->last_victim[p.gc_mode] %=
+				(MAIN_SECS(sbi) * sbi->segs_per_sec);
 			break;
 		}
 	}
@@ -1361,3 +1363,179 @@ void f2fs_build_gc_manager(struct f2fs_sb_info *sbi)
 		SIT_I(sbi)->last_victim[ALLOC_NEXT] =
 				GET_SEGNO(sbi, FDEV(0).end_blk) + 1;
 }
+
+static int free_segment_range(struct f2fs_sb_info *sbi, unsigned int start,
+							unsigned int end)
+{
+	int type;
+	unsigned int segno, next_inuse;
+	int err = 0;
+
+	/* 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);
+
+	/* do GC to move out valid blocks in the range */
+	for (segno = start; segno <= end; segno += sbi->segs_per_sec) {
+		struct gc_inode_list gc_list = {
+			.ilist = LIST_HEAD_INIT(gc_list.ilist),
+			.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);
+		put_gc_inode(&gc_list);
+
+		if (get_valid_blocks(sbi, segno, true))
+			return -EAGAIN;
+	}
+
+	err = f2fs_sync_fs(sbi->sb, 1);
+	if (err)
+		return err;
+
+	next_inuse = find_next_inuse(FREE_I(sbi), end + 1, start);
+	if (next_inuse <= end) {
+		f2fs_msg(sbi->sb, KERN_ERR,
+			"segno %u should be free but still inuse!", next_inuse);
+		f2fs_bug_on(sbi, 1);
+	}
+	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 segs = secs * sbi->segs_per_sec;
+
+	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);
+}
+
+static void update_fs_metadata(struct f2fs_sb_info *sbi, int secs)
+{
+	int segs = secs * sbi->segs_per_sec;
+	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;
+	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);
+}
+
+int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count)
+{
+	__u64 old_block_count, shrunk_blocks;
+	unsigned int secs;
+	int gc_mode, gc_type;
+	int err = 0;
+	__u32 rem;
+
+	old_block_count = le64_to_cpu(F2FS_RAW_SUPER(sbi)->block_count);
+	if (block_count > 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)
+		return -EINVAL;
+
+	if (block_count == old_block_count)
+		return 0;
+
+	if (is_sbi_flag_set(sbi, SBI_NEED_FSCK)) {
+		f2fs_msg(sbi->sb, KERN_ERR,
+			"Should run fsck to repair first.");
+		return -EINVAL;
+	}
+
+	if (test_opt(sbi, DISABLE_CHECKPOINT)) {
+		f2fs_msg(sbi->sb, KERN_ERR,
+			"Checkpoint should be enabled.");
+		return -EINVAL;
+	}
+
+	freeze_bdev(sbi->sb->s_bdev);
+
+	shrunk_blocks = old_block_count - block_count;
+	secs = div_u64(shrunk_blocks, BLKS_PER_SEC(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;
+	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;
+
+	update_sb_metadata(sbi, -secs);
+
+	err = f2fs_commit_super(sbi, false);
+	if (err) {
+		update_sb_metadata(sbi, secs);
+		goto out;
+	}
+
+	update_fs_metadata(sbi, -secs);
+	clear_sbi_flag(sbi, SBI_IS_RESIZEFS);
+	err = f2fs_sync_fs(sbi->sb, 1);
+	if (err) {
+		update_fs_metadata(sbi, secs);
+		update_sb_metadata(sbi, secs);
+		f2fs_commit_super(sbi, false);
+	}
+out:
+	if (err) {
+		set_sbi_flag(sbi, SBI_NEED_FSCK);
+		f2fs_msg(sbi->sb, KERN_ERR,
+				"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);
+	}
+	clear_sbi_flag(sbi, SBI_IS_RESIZEFS);
+	mutex_unlock(&sbi->resize_mutex);
+	thaw_bdev(sbi->sb->s_bdev, sbi->sb);
+	return err;
+}