* [PATCH 2/2] UBI: support per-device wear-leveling threshold
2026-07-23 2:02 [PATCH 0/2] UBI: fix ubi.mtd MODULE_PARM_DESC issues and add per-device wear-leveling threshold Ran Hongyun
2026-07-23 2:02 ` [PATCH 1/2] UBI: fix two issues in the ubi.mtd MODULE_PARM_DESC Ran Hongyun
@ 2026-07-23 2:02 ` Ran Hongyun
2026-07-23 3:01 ` Zhihao Cheng
` (2 more replies)
1 sibling, 3 replies; 9+ messages in thread
From: Ran Hongyun @ 2026-07-23 2:02 UTC (permalink / raw)
To: richard, miquel.raynal, vigneshr, vapier, artem.bityutskiy
Cc: linux-mtd, linux-kernel, chengzhihao1, yangerkun, yi.zhang
The UBI wear-leveling threshold (CONFIG_MTD_UBI_WL_THRESHOLD) is
currently a compile-time constant shared by all UBI devices. When a
single kernel image must support multiple NAND flashes with different
erase lifetimes, one global threshold cannot suit all devices.
Add a per-device configurable wl_threshold parameter:
- UAPI: add __s32 wl_threshold to struct ubi_attach_req, carved
from the existing padding.
- Module parameter: ubi.mtd gains a new optional token
"wl_threshold":
ubi.mtd=0,0,0,0,0,0,256 ubi.mtd=1,0,0,0,0,0,4096
0 means "use the kernel default", the accepted range is
2-65536.
Signed-off-by: Ran Hongyun <ranhongyun1@huawei.com>
---
drivers/mtd/ubi/build.c | 43 ++++++++++++++++++++++++++++++------
drivers/mtd/ubi/cdev.c | 2 +-
drivers/mtd/ubi/fastmap-wl.c | 8 +++----
drivers/mtd/ubi/ubi.h | 21 +++++++++++++++++-
drivers/mtd/ubi/wl.c | 39 ++++++++------------------------
include/uapi/mtd/ubi-user.h | 10 ++++++++-
6 files changed, 79 insertions(+), 44 deletions(-)
diff --git a/drivers/mtd/ubi/build.c b/drivers/mtd/ubi/build.c
index 8e5f178168cb..ed0216f70220 100644
--- a/drivers/mtd/ubi/build.c
+++ b/drivers/mtd/ubi/build.c
@@ -36,7 +36,7 @@
#define MTD_PARAM_LEN_MAX 64
/* Maximum number of comma-separated items in the 'mtd=' parameter */
-#define MTD_PARAM_MAX_COUNT 6
+#define MTD_PARAM_MAX_COUNT 7
/* Maximum value for the number of bad PEBs per 1024 PEBs */
#define MAX_MTD_UBI_BEB_LIMIT 768
@@ -56,6 +56,7 @@
* @max_beb_per1024: maximum expected number of bad PEBs per 1024 PEBs
* @enable_fm: enable fastmap when value is non-zero
* @need_resv_pool: reserve pool->max_size pebs when value is none-zero
+ * @wl_threshold: wear-leveling threshold, 0 means use CONFIG_MTD_UBI_WL_THRESHOLD
*/
struct mtd_dev_param {
char name[MTD_PARAM_LEN_MAX];
@@ -64,6 +65,7 @@ struct mtd_dev_param {
int max_beb_per1024;
int enable_fm;
int need_resv_pool;
+ int wl_threshold;
};
/* Numbers of elements set in the @mtd_dev_param array */
@@ -832,6 +834,8 @@ static int autoresize(struct ubi_device *ubi, int vol_id)
* @max_beb_per1024: maximum expected number of bad PEB per 1024 PEBs
* @disable_fm: whether disable fastmap
* @need_resv_pool: whether reserve pebs to fill fm_pool
+ * @wl_threshold: wear-leveling threshold for this UBI device; 0 means use
+ * %CONFIG_MTD_UBI_WL_THRESHOLD; accepted range is 2-65536
*
* This function attaches MTD device @mtd_dev to UBI and assign @ubi_num number
* to the newly created UBI device, unless @ubi_num is %UBI_DEV_NUM_AUTO, in
@@ -848,7 +852,7 @@ static int autoresize(struct ubi_device *ubi, int vol_id)
*/
int ubi_attach_mtd_dev(struct mtd_info *mtd, int ubi_num,
int vid_hdr_offset, int max_beb_per1024, bool disable_fm,
- bool need_resv_pool)
+ bool need_resv_pool, int wl_threshold)
{
struct ubi_device *ubi;
int i, err;
@@ -859,6 +863,15 @@ int ubi_attach_mtd_dev(struct mtd_info *mtd, int ubi_num,
if (!max_beb_per1024)
max_beb_per1024 = CONFIG_MTD_UBI_BEB_LIMIT;
+ if (!wl_threshold)
+ wl_threshold = CONFIG_MTD_UBI_WL_THRESHOLD;
+
+ if (wl_threshold < 2 || wl_threshold > 65536) {
+ pr_err("ubi: bad wear-leveling threshold %d\n",
+ wl_threshold);
+ return -EINVAL;
+ }
+
/*
* Check if we already have the same MTD device attached.
*
@@ -944,6 +957,8 @@ int ubi_attach_mtd_dev(struct mtd_info *mtd, int ubi_num,
ubi->ubi_num = ubi_num;
ubi->vid_hdr_offset = vid_hdr_offset;
ubi->autoresize_vol_id = -1;
+ ubi->wl_threshold = wl_threshold;
+ ubi->wl_free_max_diff = wl_threshold * 2;
#ifdef CONFIG_MTD_UBI_FASTMAP
ubi->fm_pool.used = ubi->fm_pool.size = 0;
@@ -1044,7 +1059,7 @@ int ubi_attach_mtd_dev(struct mtd_info *mtd, int ubi_num,
ubi->vol_count - UBI_INT_VOL_COUNT, UBI_INT_VOL_COUNT,
ubi->vtbl_slots);
ubi_msg(ubi, "max/mean erase counter: %d/%d, WL threshold: %d, image sequence number: %u",
- ubi->max_ec, ubi->mean_ec, CONFIG_MTD_UBI_WL_THRESHOLD,
+ ubi->max_ec, ubi->mean_ec, ubi->wl_threshold,
ubi->image_seq);
ubi_msg(ubi, "available PEBs: %d, total reserved PEBs: %d, PEBs reserved for bad PEB handling: %d",
ubi->avail_pebs, ubi->rsvd_pebs, ubi->beb_rsvd_pebs);
@@ -1247,7 +1262,7 @@ static void ubi_notify_add(struct mtd_info *mtd)
/* called while holding mtd_table_mutex */
mutex_lock_nested(&ubi_devices_mutex, SINGLE_DEPTH_NESTING);
- err = ubi_attach_mtd_dev(mtd, UBI_DEV_NUM_AUTO, 0, 0, false, false);
+ err = ubi_attach_mtd_dev(mtd, UBI_DEV_NUM_AUTO, 0, 0, false, false, 0);
mutex_unlock(&ubi_devices_mutex);
if (err < 0)
__put_mtd_device(mtd);
@@ -1289,7 +1304,8 @@ static int __init ubi_init_attach(void)
err = ubi_attach_mtd_dev(mtd, p->ubi_num,
p->vid_hdr_offs, p->max_beb_per1024,
p->enable_fm == 0,
- p->need_resv_pool != 0);
+ p->need_resv_pool != 0,
+ p->wl_threshold);
mutex_unlock(&ubi_devices_mutex);
if (err < 0) {
pr_err("UBI error: cannot attach mtd%d\n",
@@ -1569,12 +1585,24 @@ static int ubi_mtd_param_parse(const char *val, const struct kernel_param *kp)
} else
p->need_resv_pool = 0;
+ token = tokens[6];
+ if (token) {
+ int err = kstrtoint(token, 10, &p->wl_threshold);
+
+ if (err) {
+ pr_err("UBI error: bad value for wl_threshold parameter: %s\n",
+ token);
+ return -EINVAL;
+ }
+ } else
+ p->wl_threshold = 0;
+
mtd_devs += 1;
return 0;
}
module_param_call(mtd, ubi_mtd_param_parse, NULL, NULL, 0400);
-MODULE_PARM_DESC(mtd, "MTD devices to attach. Parameter format: mtd=<name|num|path>[,<vid_hdr_offs>[,max_beb_per1024[,ubi_num[,enable_fm[,need_resv_pool]]]]].\n"
+MODULE_PARM_DESC(mtd, "MTD devices to attach. Parameter format: mtd=<name|num|path>[,<vid_hdr_offs>[,max_beb_per1024[,ubi_num[,enable_fm[,need_resv_pool[,wl_threshold]]]]]].\n"
"Multiple \"mtd\" parameters may be specified.\n"
"MTD devices may be specified by their number, name, or path to the MTD character device node.\n"
"Optional \"vid_hdr_offs\" parameter specifies UBI VID header position to be used by UBI. (default value if 0)\n"
@@ -1589,7 +1617,8 @@ MODULE_PARM_DESC(mtd, "MTD devices to attach. Parameter format: mtd=<name|num|pa
"Example 3: mtd=/dev/mtd1,0,25 - attach MTD device /dev/mtd1 using default VID header offset and reserve 25*nand_size_in_blocks/1024 erase blocks for bad block handling.\n"
"\t(e.g. if the NAND *chipset* has 4096 PEB, 100 will be reserved for this UBI device).\n"
"Example 4: mtd=/dev/mtd1,0,0,5 - attach MTD device /dev/mtd1 to UBI 5 and using default values for the other fields.\n"
- "example 5: mtd=1,0,0,5 mtd=2,0,0,6,1 - attach MTD device /dev/mtd1 to UBI 5 and disable fastmap; attach MTD device /dev/mtd2 to UBI 6 and enable fastmap.(only works when fastmap is enabled and fm_autoconvert=Y).\n");
+ "example 5: mtd=1,0,0,5 mtd=2,0,0,6,1 - attach MTD device /dev/mtd1 to UBI 5 and disable fastmap; attach MTD device /dev/mtd2 to UBI 6 and enable fastmap.(only works when fastmap is enabled and fm_autoconvert=Y).\n"
+ "Example 6: mtd=/dev/mtd0,0,0,0,0,0,256 mtd=/dev/mtd1,0,0,0,0,0,4096 - attach MTD device /dev/mtd0 with wear-leveling threshold 256, and MTD device /dev/mtd1 with threshold 4096.\n");
#ifdef CONFIG_MTD_UBI_FASTMAP
module_param(fm_autoconvert, bool, 0644);
MODULE_PARM_DESC(fm_autoconvert, "Set this parameter to enable fastmap automatically on images without a fastmap.");
diff --git a/drivers/mtd/ubi/cdev.c b/drivers/mtd/ubi/cdev.c
index fd39030dbf89..54acd8394d67 100644
--- a/drivers/mtd/ubi/cdev.c
+++ b/drivers/mtd/ubi/cdev.c
@@ -1112,7 +1112,7 @@ static long ctrl_cdev_ioctl(struct file *file, unsigned int cmd,
mutex_lock(&ubi_devices_mutex);
err = ubi_attach_mtd_dev(mtd, req.ubi_num, req.vid_hdr_offset,
req.max_beb_per1024, !!req.disable_fm,
- !!req.need_resv_pool);
+ !!req.need_resv_pool, req.wl_threshold);
mutex_unlock(&ubi_devices_mutex);
if (err < 0)
put_mtd_device(mtd);
diff --git a/drivers/mtd/ubi/fastmap-wl.c b/drivers/mtd/ubi/fastmap-wl.c
index a0f750411f9d..e24bb6b18e90 100644
--- a/drivers/mtd/ubi/fastmap-wl.c
+++ b/drivers/mtd/ubi/fastmap-wl.c
@@ -238,7 +238,7 @@ void ubi_refill_pools_and_lock(struct ubi_device *ubi)
if (left_free <= 0)
break;
- e = find_wl_entry(ubi, &ubi->free, WL_FREE_MAX_DIFF,
+ e = find_wl_entry(ubi, &ubi->free, ubi->wl_free_max_diff,
!can_fill_pools(ubi, left_free));
self_check_in_wl_tree(ubi, e, &ubi->free);
rb_erase(&e->u.rb, &ubi->free);
@@ -392,18 +392,18 @@ static bool need_wear_leveling(struct ubi_device *ubi)
if (!e) {
if (!ubi->free.rb_node)
return false;
- e = find_wl_entry(ubi, &ubi->free, WL_FREE_MAX_DIFF, 0);
+ e = find_wl_entry(ubi, &ubi->free, ubi->wl_free_max_diff, 0);
ec = e->ec;
} else {
ec = e->ec;
if (ubi->free.rb_node) {
- e = find_wl_entry(ubi, &ubi->free, WL_FREE_MAX_DIFF, 0);
+ e = find_wl_entry(ubi, &ubi->free, ubi->wl_free_max_diff, 0);
ec = max(ec, e->ec);
}
}
e = rb_entry(rb_first(&ubi->used), struct ubi_wl_entry, u.rb);
- return ec - e->ec >= UBI_WL_THRESHOLD;
+ return ec - e->ec >= ubi->wl_threshold;
}
/* get_peb_for_wl - returns a PEB to be used internally by the WL sub-system.
diff --git a/drivers/mtd/ubi/ubi.h b/drivers/mtd/ubi/ubi.h
index af466cd83ae0..62026931d05e 100644
--- a/drivers/mtd/ubi/ubi.h
+++ b/drivers/mtd/ubi/ubi.h
@@ -516,6 +516,22 @@ struct ubi_debug_info {
* @bgt_thread: background thread description object
* @thread_enabled: if the background thread is enabled
* @bgt_name: background thread name
+ * @wl_threshold: Maximum difference between two erase counters. If this
+ * threshold is exceeded, the WL sub-system starts moving
+ * data from used physical eraseblocks with low erase
+ * counter to free physical eraseblocks with high erase counter.
+ * @wl_free_max_diff: When a physical eraseblock is moved, the WL sub-system
+ * has to pick the target physical eraseblock to move to.
+ * The simplest way would be just to pick the one with the
+ * highest erase counter. But in certain workloads this
+ * could lead to an unlimited wear of one or few physical
+ * eraseblock. Indeed, imagine a situation when the picked
+ * physical eraseblock is constantly erased after the
+ * data is written to it. So, we have a constant which
+ * limits the highest erase counter of the free physical
+ * eraseblock to pick. Namely, the WL sub-system does not
+ * pick eraseblocks with erase counter greater than the
+ * lowest erase counter plus @wl_free_max_diff.
*
* @flash_size: underlying MTD device size (in bytes)
* @peb_count: count of physical eraseblocks on the MTD device
@@ -623,6 +639,8 @@ struct ubi_device {
struct task_struct *bgt_thread;
int thread_enabled;
char bgt_name[sizeof(UBI_BGT_NAME_PATTERN)+2];
+ int wl_threshold;
+ int wl_free_max_diff;
/* I/O sub-system's stuff */
long long flash_size;
@@ -938,7 +956,8 @@ int ubi_io_write_vid_hdr(struct ubi_device *ubi, int pnum,
/* build.c */
int ubi_attach_mtd_dev(struct mtd_info *mtd, int ubi_num,
int vid_hdr_offset, int max_beb_per1024,
- bool disable_fm, bool need_resv_pool);
+ bool disable_fm, bool need_resv_pool,
+ int wl_threshold);
int ubi_detach_mtd_dev(int ubi_num, int anyway);
struct ubi_device *ubi_get_device(int ubi_num);
void ubi_put_device(struct ubi_device *ubi);
diff --git a/drivers/mtd/ubi/wl.c b/drivers/mtd/ubi/wl.c
index e3705db8e570..4e18044a20c0 100644
--- a/drivers/mtd/ubi/wl.c
+++ b/drivers/mtd/ubi/wl.c
@@ -95,27 +95,6 @@
/* Number of physical eraseblocks reserved for wear-leveling purposes */
#define WL_RESERVED_PEBS 1
-/*
- * Maximum difference between two erase counters. If this threshold is
- * exceeded, the WL sub-system starts moving data from used physical
- * eraseblocks with low erase counter to free physical eraseblocks with high
- * erase counter.
- */
-#define UBI_WL_THRESHOLD CONFIG_MTD_UBI_WL_THRESHOLD
-
-/*
- * When a physical eraseblock is moved, the WL sub-system has to pick the target
- * physical eraseblock to move to. The simplest way would be just to pick the
- * one with the highest erase counter. But in certain workloads this could lead
- * to an unlimited wear of one or few physical eraseblock. Indeed, imagine a
- * situation when the picked physical eraseblock is constantly erased after the
- * data is written to it. So, we have a constant which limits the highest erase
- * counter of the free physical eraseblock to pick. Namely, the WL sub-system
- * does not pick eraseblocks with erase counter greater than the lowest erase
- * counter plus %WL_FREE_MAX_DIFF.
- */
-#define WL_FREE_MAX_DIFF (2*UBI_WL_THRESHOLD)
-
/*
* Maximum number of consecutive background thread failures which is enough to
* switch to read-only mode.
@@ -358,7 +337,7 @@ static struct ubi_wl_entry *find_wl_entry(struct ubi_device *ubi,
*
* This function looks for a wear leveling entry with medium erase counter,
* but not greater or equivalent than the lowest erase counter plus
- * %WL_FREE_MAX_DIFF/2.
+ * @ubi->wl_free_max_diff/2.
*/
static struct ubi_wl_entry *find_mean_wl_entry(struct ubi_device *ubi,
struct rb_root *root)
@@ -368,7 +347,7 @@ static struct ubi_wl_entry *find_mean_wl_entry(struct ubi_device *ubi,
first = rb_entry(rb_first(root), struct ubi_wl_entry, u.rb);
last = rb_entry(rb_last(root), struct ubi_wl_entry, u.rb);
- if (last->ec - first->ec < WL_FREE_MAX_DIFF) {
+ if (last->ec - first->ec < ubi->wl_free_max_diff) {
e = rb_entry(root->rb_node, struct ubi_wl_entry, u.rb);
/*
@@ -379,7 +358,7 @@ static struct ubi_wl_entry *find_mean_wl_entry(struct ubi_device *ubi,
*/
e = may_reserve_for_fm(ubi, e, root);
} else
- e = find_wl_entry(ubi, root, WL_FREE_MAX_DIFF/2, 0);
+ e = find_wl_entry(ubi, root, ubi->wl_free_max_diff/2, 0);
return e;
}
@@ -706,7 +685,7 @@ static int wear_leveling_worker(struct ubi_device *ubi, struct ubi_work *wrk,
#ifdef CONFIG_MTD_UBI_FASTMAP
e1 = find_anchor_wl_entry(&ubi->used);
if (e1 && ubi->fm_anchor &&
- (ubi->fm_anchor->ec - e1->ec >= UBI_WL_THRESHOLD)) {
+ (ubi->fm_anchor->ec - e1->ec >= ubi->wl_threshold)) {
ubi->fm_do_produce_anchor = 1;
/*
* fm_anchor is no longer considered a good anchor.
@@ -743,7 +722,7 @@ static int wear_leveling_worker(struct ubi_device *ubi, struct ubi_work *wrk,
if (!e2)
goto out_cancel;
- if (!(e2->ec - e1->ec >= UBI_WL_THRESHOLD)) {
+ if (!(e2->ec - e1->ec >= ubi->wl_threshold)) {
dbg_wl("no WL needed: min used EC %d, max free EC %d",
e1->ec, e2->ec);
@@ -1056,12 +1035,12 @@ static int ensure_wear_leveling(struct ubi_device *ubi, int nested)
* We schedule wear-leveling only if the difference between the
* lowest erase counter of used physical eraseblocks and a high
* erase counter of free physical eraseblocks is greater than
- * %UBI_WL_THRESHOLD.
+ * @ubi->wl_threshold.
*/
e1 = rb_entry(rb_first(&ubi->used), struct ubi_wl_entry, u.rb);
- e2 = find_wl_entry(ubi, &ubi->free, WL_FREE_MAX_DIFF, 0);
+ e2 = find_wl_entry(ubi, &ubi->free, ubi->wl_free_max_diff, 0);
- if (!(e2->ec - e1->ec >= UBI_WL_THRESHOLD))
+ if (!(e2->ec - e1->ec >= ubi->wl_threshold))
goto out_unlock;
#endif
dbg_wl("schedule wear-leveling");
@@ -2090,7 +2069,7 @@ static struct ubi_wl_entry *get_peb_for_wl(struct ubi_device *ubi)
{
struct ubi_wl_entry *e;
- e = find_wl_entry(ubi, &ubi->free, WL_FREE_MAX_DIFF, 0);
+ e = find_wl_entry(ubi, &ubi->free, ubi->wl_free_max_diff, 0);
self_check_in_wl_tree(ubi, e, &ubi->free);
ubi->free_count--;
ubi_assert(ubi->free_count >= 0);
diff --git a/include/uapi/mtd/ubi-user.h b/include/uapi/mtd/ubi-user.h
index aa872a41ffb9..3538e11b5175 100644
--- a/include/uapi/mtd/ubi-user.h
+++ b/include/uapi/mtd/ubi-user.h
@@ -289,6 +289,13 @@ enum {
* If @disable_fm is not zero, ubi doesn't create new fastmap even the module
* param 'fm_autoconvert' is set, and existed old fastmap will be destroyed
* after doing full scanning.
+ *
+ * The @wl_threshold defines the maximum difference between the highest and the
+ * lowest erase counter value of eraseblocks of this UBI device. When this
+ * threshold is exceeded, UBI starts performing wear leveling by means of
+ * moving data from eraseblock with low erase counter to eraseblocks with high
+ * erase counter. If @wl_threshold is zero, the default kernel value of
+ * %CONFIG_MTD_UBI_WL_THRESHOLD is used. The accepted range is 2-65536.
*/
struct ubi_attach_req {
__s32 ubi_num;
@@ -297,7 +304,8 @@ struct ubi_attach_req {
__s16 max_beb_per1024;
__s8 disable_fm;
__s8 need_resv_pool;
- __s8 padding[8];
+ __s32 wl_threshold;
+ __s8 padding[4];
};
/*
--
2.52.0
^ permalink raw reply related [flat|nested] 9+ messages in thread