f2fs: add write priority option based on zone UFS

Currently, we are using a mix of traditional UFS and zone UFS to support
some functionalities that cannot be achieved on zone UFS alone. However,
there are some issues with this approach. There exists a significant
performance difference between traditional UFS and zone UFS. Under normal
usage, we prioritize writes to zone UFS. However, in critical conditions
(such as when the entire UFS is almost full), we cannot determine whether
data will be written to traditional UFS or zone UFS. This can lead to
significant performance fluctuations, which is not conducive to
development and testing. To address this, we have added an option
zlu_io_enable under sys with the following three modes:
1) zlu_io_enable == 0:Normal mode, prioritize writing to zone UFS;
2) zlu_io_enable == 1:Zone UFS only mode, only allow writing to zone UFS;
3) zlu_io_enable == 2:Traditional UFS priority mode, prioritize writing to
traditional UFS.

Signed-off-by: Liao Yuanhong <[email protected]>
Signed-off-by: Wu Bo <[email protected]>
Reviewed-by: Chao Yu <[email protected]>
Signed-off-by: Jaegeuk Kim <[email protected]>

Documentation/ABI/testing/sysfs-fs-f2fs

@@ -763,3 +763,17 @@ Date:		November 2023
 Contact:	"Chao Yu" <[email protected]>
 Description:	It controls to enable/disable IO aware feature for background discard.
 		By default, the value is 1 which indicates IO aware is on.
+
+What:		/sys/fs/f2fs/<disk>/blkzone_alloc_policy
+Date:		July 2024
+Contact:	"Yuanhong Liao" <[email protected]>
+Description:	The zone UFS we are currently using consists of two parts:
+		conventional zones and sequential zones. It can be used to control which part
+		to prioritize for writes, with a default value of 0.
+
+		========================  =========================================
+		value					  description
+		blkzone_alloc_policy = 0  Prioritize writing to sequential zones
+		blkzone_alloc_policy = 1  Only allow writing to sequential zones
+		blkzone_alloc_policy = 2  Prioritize writing to conventional zones
+		========================  =========================================

fs/f2fs/f2fs.h

@@ -134,6 +134,12 @@ typedef u32 nid_t;
 
 #define COMPRESS_EXT_NUM		16
 
+enum blkzone_allocation_policy {
+	BLKZONE_ALLOC_PRIOR_SEQ,	/* Prioritize writing to sequential zones */
+	BLKZONE_ALLOC_ONLY_SEQ,		/* Only allow writing to sequential zones */
+	BLKZONE_ALLOC_PRIOR_CONV,	/* Prioritize writing to conventional zones */
+};
+
 /*
  * An implementation of an rwsem that is explicitly unfair to readers. This
  * prevents priority inversion when a low-priority reader acquires the read lock
@@ -1559,6 +1565,8 @@ struct f2fs_sb_info {
 #ifdef CONFIG_BLK_DEV_ZONED
 	unsigned int blocks_per_blkz;		/* F2FS blocks per zone */
 	unsigned int max_open_zones;		/* max open zone resources of the zoned device */
+	/* For adjust the priority writing position of data in zone UFS */
+	unsigned int blkzone_alloc_policy;
 #endif
 
 	/* for node-related operations */

fs/f2fs/segment.c

@@ -2686,17 +2686,40 @@ static int get_new_segment(struct f2fs_sb_info *sbi,
 			goto got_it;
 	}
 
+#ifdef CONFIG_BLK_DEV_ZONED
 	/*
 	 * If we format f2fs on zoned storage, let's try to get pinned sections
 	 * from beginning of the storage, which should be a conventional one.
 	 */
 	if (f2fs_sb_has_blkzoned(sbi)) {
-		segno = pinning ? 0 : max(first_zoned_segno(sbi), *newseg);
+		/* Prioritize writing to conventional zones */
+		if (sbi->blkzone_alloc_policy == BLKZONE_ALLOC_PRIOR_CONV || pinning)
+			segno = 0;
+		else
+			segno = max(first_zoned_segno(sbi), *newseg);
 		hint = GET_SEC_FROM_SEG(sbi, segno);
 	}
+#endif
 
 find_other_zone:
 	secno = find_next_zero_bit(free_i->free_secmap, MAIN_SECS(sbi), hint);
+
+#ifdef CONFIG_BLK_DEV_ZONED
+	if (secno >= MAIN_SECS(sbi) && f2fs_sb_has_blkzoned(sbi)) {
+		/* Write only to sequential zones */
+		if (sbi->blkzone_alloc_policy == BLKZONE_ALLOC_ONLY_SEQ) {
+			hint = GET_SEC_FROM_SEG(sbi, first_zoned_segno(sbi));
+			secno = find_next_zero_bit(free_i->free_secmap, MAIN_SECS(sbi), hint);
+		} else
+			secno = find_first_zero_bit(free_i->free_secmap,
+								MAIN_SECS(sbi));
+		if (secno >= MAIN_SECS(sbi)) {
+			ret = -ENOSPC;
+			goto out_unlock;
+		}
+	}
+#endif
+
 	if (secno >= MAIN_SECS(sbi)) {
 		secno = find_first_zero_bit(free_i->free_secmap,
 							MAIN_SECS(sbi));

fs/f2fs/super.c

@@ -4219,6 +4219,7 @@ static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
 	sbi->aligned_blksize = true;
 #ifdef CONFIG_BLK_DEV_ZONED
 	sbi->max_open_zones = UINT_MAX;
+	sbi->blkzone_alloc_policy = BLKZONE_ALLOC_PRIOR_SEQ;
 #endif
 
 	for (i = 0; i < max_devices; i++) {

fs/f2fs/sysfs.c

@@ -627,6 +627,15 @@ static ssize_t __sbi_store(struct f2fs_attr *a,
 	}
 #endif
 
+#ifdef CONFIG_BLK_DEV_ZONED
+	if (!strcmp(a->attr.name, "blkzone_alloc_policy")) {
+		if (t < BLKZONE_ALLOC_PRIOR_SEQ || t > BLKZONE_ALLOC_PRIOR_CONV)
+			return -EINVAL;
+		sbi->blkzone_alloc_policy = t;
+		return count;
+	}
+#endif
+
 #ifdef CONFIG_F2FS_FS_COMPRESSION
 	if (!strcmp(a->attr.name, "compr_written_block") ||
 		!strcmp(a->attr.name, "compr_saved_block")) {
@@ -1033,6 +1042,7 @@ F2FS_SBI_GENERAL_RW_ATTR(warm_data_age_threshold);
 F2FS_SBI_GENERAL_RW_ATTR(last_age_weight);
 #ifdef CONFIG_BLK_DEV_ZONED
 F2FS_SBI_GENERAL_RO_ATTR(unusable_blocks_per_sec);
+F2FS_SBI_GENERAL_RW_ATTR(blkzone_alloc_policy);
 #endif
 
 /* STAT_INFO ATTR */
@@ -1187,6 +1197,7 @@ static struct attribute *f2fs_attrs[] = {
 #endif
 #ifdef CONFIG_BLK_DEV_ZONED
 	ATTR_LIST(unusable_blocks_per_sec),
+	ATTR_LIST(blkzone_alloc_policy),
 #endif
 #ifdef CONFIG_F2FS_FS_COMPRESSION
 	ATTR_LIST(compr_written_block),