[RFCv3 0/3] mtd: Support slc-mode for NTC CHIP

[RFCv3 0/3] mtd: Support slc-mode for NTC CHIP

[Chris Morgan]

From: Chris Morgan <macromorgan@hotmail.com>

Add support for slc-mode implemented in Linux for the Toshiba
TC58TEG5DCLTA00 NAND and Hynix H27UCG8T2ETR NAND flash found on the NTC
CHIP. This requires the addition of a paired-pages scheme, a new
parameter for MTD partitions of slc-mode, and setting the correct
paired-pages scheme for the TC58TEG5DCLTA00  and H27UCG8T2ETR flash
chips. I will likely need some help making sure this is consistent with
the quality expected of upstream.

Changes since V2:
 - Copied upstream Linux implementation of mtd_erase to fix an issue
   with creating new ubi partitions.
 - Implemented paired page scheme and added support for Hynix flash
   chip. Based on a cursory reading of the datasheet it appears to use
   the same pairing scheme as the Toshiba chip.

Changes since V1:

 - Updated mtd_read and mtd_write to match upstream Linux.
 - Additional mtd_get_master to match upstream Linux.
 - Removed notes about ubifs not working, because it is now.

Signed-off-by: Chris Morgan <macromorgan@hotmail.com>

Chris Morgan (4):
  mtd: Add support for Linux slc-mode for MLC NAND
  mtd: Add pairing info for Toshiba TC58TEG5DCLTA00 NAND
  mtd: Add pairing info for Hynix H27UCG8T2ETR NAND
  DO NOT UPSTREAM

 arch/arm/dts/sun5i-r8-chip.dts   |  39 +++
 drivers/mtd/mtdcore.c            | 564 ++++++++++++++++++++++++-------
 drivers/mtd/mtdcore.h            |   1 -
 drivers/mtd/mtdpart.c            |   7 +
 drivers/mtd/nand/raw/nand_base.c |  67 ++++
 include/configs/sunxi-common.h   |   2 +
 include/linux/mtd/mtd.h          | 190 ++++++++++-
 include/linux/mtd/partitions.h   |  24 +-
 include/mtd/mtd-abi.h            |   1 +
 9 files changed, 770 insertions(+), 125 deletions(-)

-- 
2.30.2

[RFCv3 1/3] mtd: Add support for Linux slc-mode for MLC NAND

[Chris Morgan]

From: Chris Morgan <macromorgan@hotmail.com>

Add support for setting the slc-mode parameter for a given MLC NAND
device. This allows a given device to write only to the first set
of paried pages to help reduce data corruption and increase reliability
of the NAND.

I haven't tested these patches extensively yet, but thus far I have
been able to boot from a ubifs partition from a vmlinuz, initramfs,
devicetree, and boot.scr from said partition. I have also been able
to create a new empty partition from within U-Boot.

As I am unfamiliar with NAND and have limited experience with the MTD
subsystems, please let me know what else I can do to improve the driver.

Thank you.

Signed-off-by: Chris Morgan <macromorgan@hotmail.com>
---
 drivers/mtd/mtdcore.c            | 564 ++++++++++++++++++++++++-------
 drivers/mtd/mtdcore.h            |   1 -
 drivers/mtd/mtdpart.c            |   7 +
 drivers/mtd/nand/raw/nand_base.c |  50 +++
 include/linux/mtd/mtd.h          | 190 ++++++++++-
 include/linux/mtd/partitions.h   |  24 +-
 include/mtd/mtd-abi.h            |   1 +
 7 files changed, 712 insertions(+), 125 deletions(-)

diff --git a/drivers/mtd/mtdcore.c b/drivers/mtd/mtdcore.c
index 582129d0df..efb749cdf2 100644
--- a/drivers/mtd/mtdcore.c
+++ b/drivers/mtd/mtdcore.c
@@ -397,6 +397,114 @@ static struct device_type mtd_devtype = {
 };
 #endif
 
+/**
+ * mtd_wunit_to_pairing_info - get pairing information of a wunit
+ * @mtd: pointer to new MTD device info structure
+ * @wunit: write unit we are interested in
+ * @info: returned pairing information
+ *
+ * Retrieve pairing information associated to the wunit.
+ * This is mainly useful when dealing with MLC/TLC NANDs where pages can be
+ * paired together, and where programming a page may influence the page it is
+ * paired with.
+ * The notion of page is replaced by the term wunit (write-unit) to stay
+ * consistent with the ->writesize field.
+ *
+ * The @wunit argument can be extracted from an absolute offset using
+ * mtd_offset_to_wunit(). @info is filled with the pairing information attached
+ * to @wunit.
+ *
+ * From the pairing info the MTD user can find all the wunits paired with
+ * @wunit using the following loop:
+ *
+ * for (i = 0; i < mtd_pairing_groups(mtd); i++) {
+ *	info.pair = i;
+ *	mtd_pairing_info_to_wunit(mtd, &info);
+ *	...
+ * }
+ */
+int mtd_wunit_to_pairing_info(struct mtd_info *mtd, int wunit,
+			      struct mtd_pairing_info *info)
+{
+	struct mtd_info *master = mtd_get_master(mtd);
+	int npairs = mtd_wunit_per_eb(master) / mtd_pairing_groups(master);
+
+	if (wunit < 0 || wunit >= npairs)
+		return -EINVAL;
+
+	if (master->pairing && master->pairing->get_info)
+		return master->pairing->get_info(master, wunit, info);
+
+	info->group = 0;
+	info->pair = wunit;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mtd_wunit_to_pairing_info);
+
+/**
+ * mtd_pairing_info_to_wunit - get wunit from pairing information
+ * @mtd: pointer to new MTD device info structure
+ * @info: pairing information struct
+ *
+ * Returns a positive number representing the wunit associated to the info
+ * struct, or a negative error code.
+ *
+ * This is the reverse of mtd_wunit_to_pairing_info(), and can help one to
+ * iterate over all wunits of a given pair (see mtd_wunit_to_pairing_info()
+ * doc).
+ *
+ * It can also be used to only program the first page of each pair (i.e.
+ * page attached to group 0), which allows one to use an MLC NAND in
+ * software-emulated SLC mode:
+ *
+ * info.group = 0;
+ * npairs = mtd_wunit_per_eb(mtd) / mtd_pairing_groups(mtd);
+ * for (info.pair = 0; info.pair < npairs; info.pair++) {
+ *	wunit = mtd_pairing_info_to_wunit(mtd, &info);
+ *	mtd_write(mtd, mtd_wunit_to_offset(mtd, blkoffs, wunit),
+ *		  mtd->writesize, &retlen, buf + (i * mtd->writesize));
+ * }
+ */
+int mtd_pairing_info_to_wunit(struct mtd_info *mtd,
+			      const struct mtd_pairing_info *info)
+{
+	struct mtd_info *master = mtd_get_master(mtd);
+	int ngroups = mtd_pairing_groups(master);
+	int npairs = mtd_wunit_per_eb(master) / ngroups;
+
+	if (!info || info->pair < 0 || info->pair >= npairs ||
+	    info->group < 0 || info->group >= ngroups)
+		return -EINVAL;
+
+	if (master->pairing && master->pairing->get_wunit)
+		return mtd->pairing->get_wunit(master, info);
+
+	return info->pair;
+}
+EXPORT_SYMBOL_GPL(mtd_pairing_info_to_wunit);
+
+/**
+ * mtd_pairing_groups - get the number of pairing groups
+ * @mtd: pointer to new MTD device info structure
+ *
+ * Returns the number of pairing groups.
+ *
+ * This number is usually equal to the number of bits exposed by a single
+ * cell, and can be used in conjunction with mtd_pairing_info_to_wunit()
+ * to iterate over all pages of a given pair.
+ */
+int mtd_pairing_groups(struct mtd_info *mtd)
+{
+	struct mtd_info *master = mtd_get_master(mtd);
+
+	if (!master->pairing || !master->pairing->ngroups)
+		return 1;
+
+	return master->pairing->ngroups;
+}
+EXPORT_SYMBOL_GPL(mtd_pairing_groups);
+
 /**
  *	add_mtd_device - register an MTD device
  *	@mtd: pointer to new MTD device info structure
@@ -412,6 +520,7 @@ int add_mtd_device(struct mtd_info *mtd)
 #ifndef __UBOOT__
 	struct mtd_notifier *not;
 #endif
+	struct mtd_info *master = mtd_get_master(mtd);
 	int i, error;
 
 #ifndef __UBOOT__
@@ -431,6 +540,20 @@ int add_mtd_device(struct mtd_info *mtd)
 #endif
 
 	BUG_ON(mtd->writesize == 0);
+
+	/*
+	 * MTD_SLC_ON_MLC_EMULATION can only be set on partitions, when the
+	 * master is an MLC NAND and has a proper pairing scheme defined.
+	 * We also reject masters that implement ->_writev() for now, because
+	 * NAND controller drivers don't implement this hook, and adding the
+	 * SLC -> MLC address/length conversion to this path is useless if we
+	 * don't have a user.
+	 */
+	if (mtd->flags & MTD_SLC_ON_MLC_EMULATION &&
+	    (!mtd_is_partition(mtd) || master->type != MTD_MLCNANDFLASH ||
+	     !master->pairing))
+		return -EINVAL;
+
 	mutex_lock(&mtd_table_mutex);
 
 	i = idr_alloc(&mtd_idr, mtd, 0, 0, GFP_KERNEL);
@@ -446,6 +569,14 @@ int add_mtd_device(struct mtd_info *mtd)
 	if (mtd->bitflip_threshold == 0)
 		mtd->bitflip_threshold = mtd->ecc_strength;
 
+	if (mtd->flags & MTD_SLC_ON_MLC_EMULATION) {
+		int ngroups = mtd_pairing_groups(master);
+
+		mtd->erasesize /= ngroups;
+		mtd->size = (u64)mtd_div_by_eb(mtd->size, master) *
+			    mtd->erasesize;
+	}
+
 	if (is_power_of_2(mtd->erasesize))
 		mtd->erasesize_shift = ffs(mtd->erasesize) - 1;
 	else
@@ -750,20 +881,28 @@ EXPORT_SYMBOL_GPL(get_mtd_device);
 
 int __get_mtd_device(struct mtd_info *mtd)
 {
+	struct mtd_info *master = mtd_get_master(mtd);
 	int err;
 
-	if (!try_module_get(mtd->owner))
+	if (!try_module_get(master->owner))
 		return -ENODEV;
 
-	if (mtd->_get_device) {
-		err = mtd->_get_device(mtd);
+	if (master->_get_device) {
+		err = master->_get_device(mtd);
 
 		if (err) {
-			module_put(mtd->owner);
+			module_put(master->owner);
 			return err;
 		}
 	}
-	mtd->usecount++;
+
+	master->usecount++;
+
+	while (mtd->parent) {
+		mtd->usecount++;
+		mtd = mtd->parent;
+	}
+
 	return 0;
 }
 EXPORT_SYMBOL_GPL(__get_mtd_device);
@@ -896,13 +1035,20 @@ EXPORT_SYMBOL_GPL(put_mtd_device);
 
 void __put_mtd_device(struct mtd_info *mtd)
 {
-	--mtd->usecount;
-	BUG_ON(mtd->usecount < 0);
+	struct mtd_info *master = mtd_get_master(mtd);
+
+	while (mtd->parent) {
+		--mtd->usecount;
+		BUG_ON(mtd->usecount < 0);
+		mtd = mtd->parent;
+	}
+
+	master->usecount--;
 
-	if (mtd->_put_device)
-		mtd->_put_device(mtd);
+	if (master->_put_device)
+		master->_put_device(master);
 
-	module_put(mtd->owner);
+	module_put(master->owner);
 }
 EXPORT_SYMBOL_GPL(__put_mtd_device);
 
@@ -915,17 +1061,47 @@ EXPORT_SYMBOL_GPL(__put_mtd_device);
  */
 int mtd_erase(struct mtd_info *mtd, struct erase_info *instr)
 {
+	struct mtd_info *master = mtd_get_master(mtd);
+	u64 mst_ofs = mtd_get_master_ofs(mtd, 0);
+	struct erase_info adjinstr;
+	int ret;
+
+	instr->fail_addr = MTD_FAIL_ADDR_UNKNOWN;
+	adjinstr = *instr;
+
+	if (!mtd->erasesize || !master->_erase)
+		return -ENOTSUPP;
+
 	if (instr->addr > mtd->size || instr->len > mtd->size - instr->addr)
 		return -EINVAL;
 	if (!(mtd->flags & MTD_WRITEABLE))
 		return -EROFS;
-	instr->fail_addr = MTD_FAIL_ADDR_UNKNOWN;
-	if (!instr->len) {
-		instr->state = MTD_ERASE_DONE;
-		mtd_erase_callback(instr);
+
+	if (!instr->len)
 		return 0;
+
+	if (mtd->flags & MTD_SLC_ON_MLC_EMULATION) {
+		adjinstr.addr = (loff_t)mtd_div_by_eb(instr->addr, mtd) *
+				master->erasesize;
+		adjinstr.len = ((u64)mtd_div_by_eb(instr->addr + instr->len, mtd) *
+				master->erasesize) -
+			       adjinstr.addr;
 	}
-	return mtd->_erase(mtd, instr);
+
+	adjinstr.addr += mst_ofs;
+
+	ret = master->_erase(master, &adjinstr);
+
+	if (adjinstr.fail_addr != MTD_FAIL_ADDR_UNKNOWN) {
+		instr->fail_addr = adjinstr.fail_addr - mst_ofs;
+		if (mtd->flags & MTD_SLC_ON_MLC_EMULATION) {
+			instr->fail_addr = mtd_div_by_eb(instr->fail_addr,
+							 master);
+			instr->fail_addr *= mtd->erasesize;
+		}
+	}
+
+	return ret;
 }
 EXPORT_SYMBOL_GPL(mtd_erase);
 
@@ -936,30 +1112,36 @@ EXPORT_SYMBOL_GPL(mtd_erase);
 int mtd_point(struct mtd_info *mtd, loff_t from, size_t len, size_t *retlen,
 	      void **virt, resource_size_t *phys)
 {
+	struct mtd_info *master = mtd_get_master(mtd);
+
 	*retlen = 0;
 	*virt = NULL;
 	if (phys)
 		*phys = 0;
-	if (!mtd->_point)
+	if (!master->_point)
 		return -EOPNOTSUPP;
 	if (from < 0 || from > mtd->size || len > mtd->size - from)
 		return -EINVAL;
 	if (!len)
 		return 0;
-	return mtd->_point(mtd, from, len, retlen, virt, phys);
+
+	from = mtd_get_master_ofs(mtd, from);
+	return master->_point(master, from, len, retlen, virt, phys);
 }
 EXPORT_SYMBOL_GPL(mtd_point);
 
 /* We probably shouldn't allow XIP if the unpoint isn't a NULL */
 int mtd_unpoint(struct mtd_info *mtd, loff_t from, size_t len)
 {
-	if (!mtd->_point)
+	struct mtd_info *master = mtd_get_master(mtd);
+
+	if (!master->_point)
 		return -EOPNOTSUPP;
 	if (from < 0 || from > mtd->size || len > mtd->size - from)
 		return -EINVAL;
 	if (!len)
 		return 0;
-	return mtd->_unpoint(mtd, from, len);
+	return master->_unpoint(master, mtd_get_master_ofs(mtd, from), len);
 }
 EXPORT_SYMBOL_GPL(mtd_unpoint);
 #endif
@@ -980,68 +1162,54 @@ unsigned long mtd_get_unmapped_area(struct mtd_info *mtd, unsigned long len,
 }
 EXPORT_SYMBOL_GPL(mtd_get_unmapped_area);
 
+static void mtd_update_ecc_stats(struct mtd_info *mtd, struct mtd_info *master,
+				 const struct mtd_ecc_stats *old_stats)
+{
+	struct mtd_ecc_stats diff;
+
+	if (master == mtd)
+		return;
+
+	diff = master->ecc_stats;
+	diff.failed -= old_stats->failed;
+	diff.corrected -= old_stats->corrected;
+
+	while (mtd->parent) {
+		mtd->ecc_stats.failed += diff.failed;
+		mtd->ecc_stats.corrected += diff.corrected;
+		mtd = mtd->parent;
+	}
+}
+
 int mtd_read(struct mtd_info *mtd, loff_t from, size_t len, size_t *retlen,
 	     u_char *buf)
 {
-	int ret_code;
-	*retlen = 0;
-	if (from < 0 || from > mtd->size || len > mtd->size - from)
-		return -EINVAL;
-	if (!len)
-		return 0;
+	struct mtd_oob_ops ops = {
+		.len = len,
+		.datbuf = buf,
+	};
+	int ret;
 
-	/*
-	 * In the absence of an error, drivers return a non-negative integer
-	 * representing the maximum number of bitflips that were corrected on
-	 * any one ecc region (if applicable; zero otherwise).
-	 */
-	if (mtd->_read) {
-		ret_code = mtd->_read(mtd, from, len, retlen, buf);
-	} else if (mtd->_read_oob) {
-		struct mtd_oob_ops ops = {
-			.len = len,
-			.datbuf = buf,
-		};
-
-		ret_code = mtd->_read_oob(mtd, from, &ops);
-		*retlen = ops.retlen;
-	} else {
-		return -ENOTSUPP;
-	}
+	ret = mtd_read_oob(mtd, from, &ops);
+	*retlen = ops.retlen;
 
-	if (unlikely(ret_code < 0))
-		return ret_code;
-	if (mtd->ecc_strength == 0)
-		return 0;	/* device lacks ecc */
-	return ret_code >= mtd->bitflip_threshold ? -EUCLEAN : 0;
+	return ret;
 }
 EXPORT_SYMBOL_GPL(mtd_read);
 
 int mtd_write(struct mtd_info *mtd, loff_t to, size_t len, size_t *retlen,
 	      const u_char *buf)
 {
-	*retlen = 0;
-	if (to < 0 || to > mtd->size || len > mtd->size - to)
-		return -EINVAL;
-	if ((!mtd->_write && !mtd->_write_oob) ||
-	    !(mtd->flags & MTD_WRITEABLE))
-		return -EROFS;
-	if (!len)
-		return 0;
-
-	if (!mtd->_write) {
-		struct mtd_oob_ops ops = {
-			.len = len,
-			.datbuf = (u8 *)buf,
-		};
-		int ret;
+	struct mtd_oob_ops ops = {
+		.len = len,
+		.datbuf = (u8 *)buf,
+	};
+	int ret;
 
-		ret = mtd->_write_oob(mtd, to, &ops);
-		*retlen = ops.retlen;
-		return ret;
-	}
+	ret = mtd_write_oob(mtd, to, &ops);
+	*retlen = ops.retlen;
 
-	return mtd->_write(mtd, to, len, retlen, buf);
+	return ret;
 }
 EXPORT_SYMBOL_GPL(mtd_write);
 
@@ -1055,8 +1223,10 @@ EXPORT_SYMBOL_GPL(mtd_write);
 int mtd_panic_write(struct mtd_info *mtd, loff_t to, size_t len, size_t *retlen,
 		    const u_char *buf)
 {
+	struct mtd_info *master = mtd_get_master(mtd);
+
 	*retlen = 0;
-	if (!mtd->_panic_write)
+	if (!master->_panic_write)
 		return -EOPNOTSUPP;
 	if (to < 0 || to > mtd->size || len > mtd->size - to)
 		return -EINVAL;
@@ -1064,7 +1234,8 @@ int mtd_panic_write(struct mtd_info *mtd, loff_t to, size_t len, size_t *retlen,
 		return -EROFS;
 	if (!len)
 		return 0;
-	return mtd->_panic_write(mtd, to, len, retlen, buf);
+	return master->_panic_write(master, mtd_get_master_ofs(mtd, to), len,
+				    retlen, buf);
 }
 EXPORT_SYMBOL_GPL(mtd_panic_write);
 
@@ -1101,9 +1272,107 @@ static int mtd_check_oob_ops(struct mtd_info *mtd, loff_t offs,
 	return 0;
 }
 
+static int mtd_read_oob_std(struct mtd_info *mtd, loff_t from,
+			    struct mtd_oob_ops *ops)
+{
+	struct mtd_info *master = mtd_get_master(mtd);
+	int ret;
+
+	from = mtd_get_master_ofs(mtd, from);
+	if (master->_read_oob)
+		ret = master->_read_oob(master, from, ops);
+	else
+		ret = master->_read(master, from, ops->len, &ops->retlen,
+				 ops->datbuf);
+
+	return ret;
+}
+
+static int mtd_write_oob_std(struct mtd_info *mtd, loff_t to,
+			     struct mtd_oob_ops *ops)
+{
+	struct mtd_info *master = mtd_get_master(mtd);
+	int ret;
+
+	to = mtd_get_master_ofs(mtd, to);
+	if (master->_write_oob)
+		ret = master->_write_oob(master, to, ops);
+	else
+		ret = master->_write(master, to, ops->len, &ops->retlen,
+				  ops->datbuf);
+
+	return ret;
+}
+
+static int mtd_io_emulated_slc(struct mtd_info *mtd, loff_t start, bool read,
+			       struct mtd_oob_ops *ops)
+{
+	struct mtd_info *master = mtd_get_master(mtd);
+	int ngroups = mtd_pairing_groups(master);
+	int npairs = mtd_wunit_per_eb(master) / ngroups;
+	struct mtd_oob_ops adjops = *ops;
+	unsigned int wunit, oobavail;
+	struct mtd_pairing_info info;
+	int max_bitflips = 0;
+	u32 ebofs, pageofs;
+	loff_t base, pos;
+
+	ebofs = mtd_mod_by_eb(start, mtd);
+	base = (loff_t)mtd_div_by_eb(start, mtd) * master->erasesize;
+	info.group = 0;
+	info.pair = mtd_div_by_ws(ebofs, mtd);
+	pageofs = mtd_mod_by_ws(ebofs, mtd);
+	oobavail = mtd_oobavail(mtd, ops);
+
+	while (ops->retlen < ops->len || ops->oobretlen < ops->ooblen) {
+		int ret;
+
+		if (info.pair >= npairs) {
+			info.pair = 0;
+			base += master->erasesize;
+		}
+
+		wunit = mtd_pairing_info_to_wunit(master, &info);
+		pos = mtd_wunit_to_offset(mtd, base, wunit);
+
+		adjops.len = ops->len - ops->retlen;
+		if (adjops.len > mtd->writesize - pageofs)
+			adjops.len = mtd->writesize - pageofs;
+
+		adjops.ooblen = ops->ooblen - ops->oobretlen;
+		if (adjops.ooblen > oobavail - adjops.ooboffs)
+			adjops.ooblen = oobavail - adjops.ooboffs;
+
+		if (read) {
+			ret = mtd_read_oob_std(mtd, pos + pageofs, &adjops);
+			if (ret > 0)
+				max_bitflips = max(max_bitflips, ret);
+		} else {
+			ret = mtd_write_oob_std(mtd, pos + pageofs, &adjops);
+		}
+
+		if (ret < 0)
+			return ret;
+
+		max_bitflips = max(max_bitflips, ret);
+		ops->retlen += adjops.retlen;
+		ops->oobretlen += adjops.oobretlen;
+		adjops.datbuf += adjops.retlen;
+		adjops.oobbuf += adjops.oobretlen;
+		adjops.ooboffs = 0;
+		pageofs = 0;
+		info.pair++;
+	}
+
+	return max_bitflips;
+}
+
 int mtd_read_oob(struct mtd_info *mtd, loff_t from, struct mtd_oob_ops *ops)
 {
+	struct mtd_info *master = mtd_get_master(mtd);
+	struct mtd_ecc_stats old_stats = master->ecc_stats;
 	int ret_code;
+
 	ops->retlen = ops->oobretlen = 0;
 
 	ret_code = mtd_check_oob_ops(mtd, from, ops);
@@ -1111,14 +1380,15 @@ int mtd_read_oob(struct mtd_info *mtd, loff_t from, struct mtd_oob_ops *ops)
 		return ret_code;
 
 	/* Check the validity of a potential fallback on mtd->_read */
-	if (!mtd->_read_oob && (!mtd->_read || ops->oobbuf))
+	if (!master->_read_oob && (!master->_read || ops->oobbuf))
 		return -EOPNOTSUPP;
 
-	if (mtd->_read_oob)
-		ret_code = mtd->_read_oob(mtd, from, ops);
+	if (mtd->flags & MTD_SLC_ON_MLC_EMULATION)
+		ret_code = mtd_io_emulated_slc(mtd, from, true, ops);
 	else
-		ret_code = mtd->_read(mtd, from, ops->len, &ops->retlen,
-				      ops->datbuf);
+		ret_code = mtd_read_oob_std(mtd, from, ops);
+
+	mtd_update_ecc_stats(mtd, master, &old_stats);
 
 	/*
 	 * In cases where ops->datbuf != NULL, mtd->_read_oob() has semantics
@@ -1137,6 +1407,7 @@ EXPORT_SYMBOL_GPL(mtd_read_oob);
 int mtd_write_oob(struct mtd_info *mtd, loff_t to,
 				struct mtd_oob_ops *ops)
 {
+	struct mtd_info *master = mtd_get_master(mtd);
 	int ret;
 
 	ops->retlen = ops->oobretlen = 0;
@@ -1149,14 +1420,14 @@ int mtd_write_oob(struct mtd_info *mtd, loff_t to,
 		return ret;
 
 	/* Check the validity of a potential fallback on mtd->_write */
-	if (!mtd->_write_oob && (!mtd->_write || ops->oobbuf))
+	if (!master->_write_oob && (!master->_write || ops->oobbuf))
 		return -EOPNOTSUPP;
 
-	if (mtd->_write_oob)
-		return mtd->_write_oob(mtd, to, ops);
-	else
-		return mtd->_write(mtd, to, ops->len, &ops->retlen,
-				   ops->datbuf);
+	if (mtd->flags & MTD_SLC_ON_MLC_EMULATION)
+		return mtd_io_emulated_slc(mtd, to, false, ops);
+
+	return mtd_write_oob_std(mtd, to, ops);
+
 }
 EXPORT_SYMBOL_GPL(mtd_write_oob);
 
@@ -1179,15 +1450,17 @@ EXPORT_SYMBOL_GPL(mtd_write_oob);
 int mtd_ooblayout_ecc(struct mtd_info *mtd, int section,
 		      struct mtd_oob_region *oobecc)
 {
+	struct mtd_info *master = mtd_get_master(mtd);
+
 	memset(oobecc, 0, sizeof(*oobecc));
 
-	if (!mtd || section < 0)
+	if (!master || section < 0)
 		return -EINVAL;
 
-	if (!mtd->ooblayout || !mtd->ooblayout->ecc)
+	if (!master->ooblayout || !master->ooblayout->ecc)
 		return -ENOTSUPP;
 
-	return mtd->ooblayout->ecc(mtd, section, oobecc);
+	return master->ooblayout->ecc(master, section, oobecc);
 }
 EXPORT_SYMBOL_GPL(mtd_ooblayout_ecc);
 
@@ -1211,15 +1484,17 @@ EXPORT_SYMBOL_GPL(mtd_ooblayout_ecc);
 int mtd_ooblayout_free(struct mtd_info *mtd, int section,
 		       struct mtd_oob_region *oobfree)
 {
+	struct mtd_info *master = mtd_get_master(mtd);
+
 	memset(oobfree, 0, sizeof(*oobfree));
 
-	if (!mtd || section < 0)
+	if (!master || section < 0)
 		return -EINVAL;
 
-	if (!mtd->ooblayout || !mtd->ooblayout->rfree)
+	if (!master->ooblayout || !master->ooblayout->rfree)
 		return -ENOTSUPP;
 
-	return mtd->ooblayout->rfree(mtd, section, oobfree);
+	return master->ooblayout->rfree(master, section, oobfree);
 }
 EXPORT_SYMBOL_GPL(mtd_ooblayout_free);
 
@@ -1528,60 +1803,69 @@ EXPORT_SYMBOL_GPL(mtd_ooblayout_count_eccbytes);
 int mtd_get_fact_prot_info(struct mtd_info *mtd, size_t len, size_t *retlen,
 			   struct otp_info *buf)
 {
-	if (!mtd->_get_fact_prot_info)
+	struct mtd_info *master = mtd_get_master(mtd);
+
+	if (!master->_get_fact_prot_info)
 		return -EOPNOTSUPP;
 	if (!len)
 		return 0;
-	return mtd->_get_fact_prot_info(mtd, len, retlen, buf);
+	return master->_get_fact_prot_info(master, len, retlen, buf);
 }
 EXPORT_SYMBOL_GPL(mtd_get_fact_prot_info);
 
 int mtd_read_fact_prot_reg(struct mtd_info *mtd, loff_t from, size_t len,
 			   size_t *retlen, u_char *buf)
 {
+	struct mtd_info *master = mtd_get_master(mtd);
+
 	*retlen = 0;
-	if (!mtd->_read_fact_prot_reg)
+	if (!master->_read_fact_prot_reg)
 		return -EOPNOTSUPP;
 	if (!len)
 		return 0;
-	return mtd->_read_fact_prot_reg(mtd, from, len, retlen, buf);
+	return master->_read_fact_prot_reg(master, from, len, retlen, buf);
 }
 EXPORT_SYMBOL_GPL(mtd_read_fact_prot_reg);
 
 int mtd_get_user_prot_info(struct mtd_info *mtd, size_t len, size_t *retlen,
 			   struct otp_info *buf)
 {
-	if (!mtd->_get_user_prot_info)
+	struct mtd_info *master = mtd_get_master(mtd);
+
+	if (!master->_get_user_prot_info)
 		return -EOPNOTSUPP;
 	if (!len)
 		return 0;
-	return mtd->_get_user_prot_info(mtd, len, retlen, buf);
+	return master->_get_user_prot_info(master, len, retlen, buf);
 }
 EXPORT_SYMBOL_GPL(mtd_get_user_prot_info);
 
 int mtd_read_user_prot_reg(struct mtd_info *mtd, loff_t from, size_t len,
 			   size_t *retlen, u_char *buf)
 {
+	struct mtd_info *master = mtd_get_master(mtd);
+
 	*retlen = 0;
-	if (!mtd->_read_user_prot_reg)
+	if (!master->_read_user_prot_reg)
 		return -EOPNOTSUPP;
 	if (!len)
 		return 0;
-	return mtd->_read_user_prot_reg(mtd, from, len, retlen, buf);
+	return master->_read_user_prot_reg(master, from, len, retlen, buf);
 }
 EXPORT_SYMBOL_GPL(mtd_read_user_prot_reg);
 
 int mtd_write_user_prot_reg(struct mtd_info *mtd, loff_t to, size_t len,
 			    size_t *retlen, u_char *buf)
 {
+	struct mtd_info *master = mtd_get_master(mtd);
 	int ret;
 
 	*retlen = 0;
-	if (!mtd->_write_user_prot_reg)
+	if (!master->_write_user_prot_reg)
 		return -EOPNOTSUPP;
 	if (!len)
 		return 0;
-	ret = mtd->_write_user_prot_reg(mtd, to, len, retlen, buf);
+	ret = master->_write_user_prot_reg(master, to, len, retlen, buf);
 	if (ret)
 		return ret;
 
@@ -1595,80 +1879,129 @@ EXPORT_SYMBOL_GPL(mtd_write_user_prot_reg);
 
 int mtd_lock_user_prot_reg(struct mtd_info *mtd, loff_t from, size_t len)
 {
-	if (!mtd->_lock_user_prot_reg)
+	struct mtd_info *master = mtd_get_master(mtd);
+
+	if (!master->_lock_user_prot_reg)
 		return -EOPNOTSUPP;
 	if (!len)
 		return 0;
-	return mtd->_lock_user_prot_reg(mtd, from, len);
+	return master->_lock_user_prot_reg(master, from, len);
 }
 EXPORT_SYMBOL_GPL(mtd_lock_user_prot_reg);
 
 /* Chip-supported device locking */
 int mtd_lock(struct mtd_info *mtd, loff_t ofs, uint64_t len)
 {
-	if (!mtd->_lock)
+	struct mtd_info *master = mtd_get_master(mtd);
+
+	if (!master->_lock)
 		return -EOPNOTSUPP;
 	if (ofs < 0 || ofs > mtd->size || len > mtd->size - ofs)
 		return -EINVAL;
 	if (!len)
 		return 0;
-	return mtd->_lock(mtd, ofs, len);
+	if (mtd->flags & MTD_SLC_ON_MLC_EMULATION) {
+		ofs = (loff_t)mtd_div_by_eb(ofs, mtd) * master->erasesize;
+		len = (u64)mtd_div_by_eb(len, mtd) * master->erasesize;
+	}
+
+	return master->_lock(master, mtd_get_master_ofs(mtd, ofs), len);
 }
 EXPORT_SYMBOL_GPL(mtd_lock);
 
 int mtd_unlock(struct mtd_info *mtd, loff_t ofs, uint64_t len)
 {
-	if (!mtd->_unlock)
+	struct mtd_info *master = mtd_get_master(mtd);
+
+	if (!master->_unlock)
 		return -EOPNOTSUPP;
 	if (ofs < 0 || ofs > mtd->size || len > mtd->size - ofs)
 		return -EINVAL;
 	if (!len)
 		return 0;
-	return mtd->_unlock(mtd, ofs, len);
+	if (mtd->flags & MTD_SLC_ON_MLC_EMULATION) {
+		ofs = (loff_t)mtd_div_by_eb(ofs, mtd) * master->erasesize;
+		len = (u64)mtd_div_by_eb(len, mtd) * master->erasesize;
+	}
+
+	return master->_unlock(master, mtd_get_master_ofs(mtd, ofs), len);
 }
 EXPORT_SYMBOL_GPL(mtd_unlock);
 
 int mtd_is_locked(struct mtd_info *mtd, loff_t ofs, uint64_t len)
 {
-	if (!mtd->_is_locked)
+	struct mtd_info *master = mtd_get_master(mtd);
+
+	if (!master->_is_locked)
 		return -EOPNOTSUPP;
 	if (ofs < 0 || ofs > mtd->size || len > mtd->size - ofs)
 		return -EINVAL;
 	if (!len)
 		return 0;
-	return mtd->_is_locked(mtd, ofs, len);
+	if (mtd->flags & MTD_SLC_ON_MLC_EMULATION) {
+		ofs = (loff_t)mtd_div_by_eb(ofs, mtd) * master->erasesize;
+		len = (u64)mtd_div_by_eb(len, mtd) * master->erasesize;
+	}
+
+	return master->_is_locked(master, mtd_get_master_ofs(mtd, ofs), len);
 }
 EXPORT_SYMBOL_GPL(mtd_is_locked);
 
 int mtd_block_isreserved(struct mtd_info *mtd, loff_t ofs)
 {
+	struct mtd_info *master = mtd_get_master(mtd);
+
 	if (ofs < 0 || ofs > mtd->size)
 		return -EINVAL;
-	if (!mtd->_block_isreserved)
+	if (!master->_block_isreserved)
 		return 0;
-	return mtd->_block_isreserved(mtd, ofs);
+	if (mtd->flags & MTD_SLC_ON_MLC_EMULATION)
+		ofs = (loff_t)mtd_div_by_eb(ofs, mtd) * master->erasesize;
+
+	return master->_block_isreserved(master, mtd_get_master_ofs(mtd, ofs));
 }
 EXPORT_SYMBOL_GPL(mtd_block_isreserved);
 
 int mtd_block_isbad(struct mtd_info *mtd, loff_t ofs)
 {
+	struct mtd_info *master = mtd_get_master(mtd);
+
 	if (ofs < 0 || ofs > mtd->size)
 		return -EINVAL;
-	if (!mtd->_block_isbad)
+	if (!master->_block_isbad)
 		return 0;
-	return mtd->_block_isbad(mtd, ofs);
+	if (mtd->flags & MTD_SLC_ON_MLC_EMULATION)
+		ofs = (loff_t)mtd_div_by_eb(ofs, mtd) * master->erasesize;
+
+	return master->_block_isbad(master, mtd_get_master_ofs(mtd, ofs));
 }
 EXPORT_SYMBOL_GPL(mtd_block_isbad);
 
 int mtd_block_markbad(struct mtd_info *mtd, loff_t ofs)
 {
-	if (!mtd->_block_markbad)
+	struct mtd_info *master = mtd_get_master(mtd);
+	int ret;
+
+	if (!master->_block_markbad)
 		return -EOPNOTSUPP;
 	if (ofs < 0 || ofs > mtd->size)
 		return -EINVAL;
 	if (!(mtd->flags & MTD_WRITEABLE))
 		return -EROFS;
-	return mtd->_block_markbad(mtd, ofs);
+
+	if (mtd->flags & MTD_SLC_ON_MLC_EMULATION)
+		ofs = (loff_t)mtd_div_by_eb(ofs, mtd) * master->erasesize;
+
+	ret = master->_block_markbad(master, mtd_get_master_ofs(mtd, ofs));
+	if (ret)
+		return ret;
+
+	while (mtd->parent) {
+		mtd->ecc_stats.badblocks++;
+		mtd = mtd->parent;
+	}
+
+	return 0;
 }
 EXPORT_SYMBOL_GPL(mtd_block_markbad);
 
@@ -1719,12 +2052,17 @@ static int default_mtd_writev(struct mtd_info *mtd, const struct kvec *vecs,
 int mtd_writev(struct mtd_info *mtd, const struct kvec *vecs,
 	       unsigned long count, loff_t to, size_t *retlen)
 {
+	struct mtd_info *master = mtd_get_master(mtd);
+
 	*retlen = 0;
 	if (!(mtd->flags & MTD_WRITEABLE))
 		return -EROFS;
-	if (!mtd->_writev)
+
+	if (!master->_writev)
 		return default_mtd_writev(mtd, vecs, count, to, retlen);
-	return mtd->_writev(mtd, vecs, count, to, retlen);
+
+	return master->_writev(master, vecs, count,
+			       mtd_get_master_ofs(mtd, to), retlen);
 }
 EXPORT_SYMBOL_GPL(mtd_writev);
 
diff --git a/drivers/mtd/mtdcore.h b/drivers/mtd/mtdcore.h
index 1d181a1045..73f8f21271 100644
--- a/drivers/mtd/mtdcore.h
+++ b/drivers/mtd/mtdcore.h
@@ -10,7 +10,6 @@ int del_mtd_device(struct mtd_info *mtd);
 int add_mtd_partitions(struct mtd_info *, const struct mtd_partition *, int);
 int del_mtd_partitions(struct mtd_info *);
 int parse_mtd_partitions(struct mtd_info *master, const char * const *types,
-			 struct mtd_partition **pparts,
 			 struct mtd_part_parser_data *data);
 
 int __init init_mtdchar(void);
diff --git a/drivers/mtd/mtdpart.c b/drivers/mtd/mtdpart.c
index aa58f722da..a686271839 100644
--- a/drivers/mtd/mtdpart.c
+++ b/drivers/mtd/mtdpart.c
@@ -148,6 +148,12 @@ static int mtd_parse_partition(const char **_mtdparts,
 		mtdparts += 2;
 	}
 
+	/* if slc is found use emulated SLC mode on this partition*/
+	if (!strncmp(mtdparts, "slc", 3)) {
+		partition->add_flags |= MTD_SLC_ON_MLC_EMULATION;
+		mtdparts += 3;
+	}
+
 	/* Check for a potential next partition definition */
 	if (*mtdparts == ',') {
 		if (partition->size == MTD_SIZE_REMAINING) {
@@ -591,6 +597,7 @@ static struct mtd_info *allocate_partition(struct mtd_info *master,
 	/* set up the MTD object for this partition */
 	slave->type = master->type;
 	slave->flags = master->flags & ~part->mask_flags;
+	slave->flags |= part->add_flags;
 	slave->size = part->size;
 	slave->writesize = master->writesize;
 	slave->writebufsize = master->writebufsize;
diff --git a/drivers/mtd/nand/raw/nand_base.c b/drivers/mtd/nand/raw/nand_base.c
index b1fd779884..9e50dce40b 100644
--- a/drivers/mtd/nand/raw/nand_base.c
+++ b/drivers/mtd/nand/raw/nand_base.c
@@ -105,6 +105,56 @@ static int nand_do_write_oob(struct mtd_info *mtd, loff_t to,
  */
 DEFINE_LED_TRIGGER(nand_led_trigger);
 
+static int nand_pairing_dist3_get_info(struct mtd_info *mtd, int page,
+				       struct mtd_pairing_info *info)
+{
+	int lastpage = (mtd->erasesize / mtd->writesize) - 1;
+	int dist = 3;
+
+	if (page == lastpage)
+		dist = 2;
+
+	if (!page || (page & 1)) {
+		info->group = 0;
+		info->pair = (page + 1) / 2;
+	} else {
+		info->group = 1;
+		info->pair = (page + 1 - dist) / 2;
+	}
+
+	return 0;
+}
+
+static int nand_pairing_dist3_get_wunit(struct mtd_info *mtd,
+					const struct mtd_pairing_info *info)
+{
+	int lastpair = ((mtd->erasesize / mtd->writesize) - 1) / 2;
+	int page = info->pair * 2;
+	int dist = 3;
+
+	if (!info->group && !info->pair)
+		return 0;
+
+	if (info->pair == lastpair && info->group)
+		dist = 2;
+
+	if (!info->group)
+		page--;
+	else if (info->pair)
+		page += dist - 1;
+
+	if (page >= mtd->erasesize / mtd->writesize)
+		return -EINVAL;
+
+	return page;
+}
+
+const struct mtd_pairing_scheme dist3_pairing_scheme = {
+	.ngroups = 2,
+	.get_info = nand_pairing_dist3_get_info,
+	.get_wunit = nand_pairing_dist3_get_wunit,
+};
+
 static int check_offs_len(struct mtd_info *mtd,
 					loff_t ofs, uint64_t len)
 {
diff --git a/include/linux/mtd/mtd.h b/include/linux/mtd/mtd.h
index 3b302fb8c3..f55e469f75 100644
--- a/include/linux/mtd/mtd.h
+++ b/include/linux/mtd/mtd.h
@@ -133,6 +133,82 @@ struct mtd_ooblayout_ops {
 		     struct mtd_oob_region *oobfree);
 };
 
+/**
+ * struct mtd_pairing_info - page pairing information
+ *
+ * @pair: pair id
+ * @group: group id
+ *
+ * The term "pair" is used here, even though TLC NANDs might group pages by 3
+ * (3 bits in a single cell). A pair should regroup all pages that are sharing
+ * the same cell. Pairs are then indexed in ascending order.
+ *
+ * @group is defining the position of a page in a given pair. It can also be
+ * seen as the bit position in the cell: page attached to bit 0 belongs to
+ * group 0, page attached to bit 1 belongs to group 1, etc.
+ *
+ * Example:
+ * The H27UCG8T2BTR-BC datasheet describes the following pairing scheme:
+ *
+ *		group-0		group-1
+ *
+ *  pair-0	page-0		page-4
+ *  pair-1	page-1		page-5
+ *  pair-2	page-2		page-8
+ *  ...
+ *  pair-127	page-251	page-255
+ *
+ *
+ * Note that the "group" and "pair" terms were extracted from Samsung and
+ * Hynix datasheets, and might be referenced under other names in other
+ * datasheets (Micron is describing this concept as "shared pages").
+ */
+struct mtd_pairing_info {
+	int pair;
+	int group;
+};
+
+/**
+ * struct mtd_pairing_scheme - page pairing scheme description
+ *
+ * @ngroups: number of groups. Should be related to the number of bits
+ *	     per cell.
+ * @get_info: converts a write-unit (page number within an erase block) into
+ *	      mtd_pairing information (pair + group). This function should
+ *	      fill the info parameter based on the wunit index or return
+ *	      -EINVAL if the wunit parameter is invalid.
+ * @get_wunit: converts pairing information into a write-unit (page) number.
+ *	       This function should return the wunit index pointed by the
+ *	       pairing information described in the info argument. It should
+ *	       return -EINVAL, if there's no wunit corresponding to the
+ *	       passed pairing information.
+ *
+ * See mtd_pairing_info documentation for a detailed explanation of the
+ * pair and group concepts.
+ *
+ * The mtd_pairing_scheme structure provides a generic solution to represent
+ * NAND page pairing scheme. Instead of exposing two big tables to do the
+ * write-unit <-> (pair + group) conversions, we ask the MTD drivers to
+ * implement the ->get_info() and ->get_wunit() functions.
+ *
+ * MTD users will then be able to query these information by using the
+ * mtd_pairing_info_to_wunit() and mtd_wunit_to_pairing_info() helpers.
+ *
+ * @ngroups is here to help MTD users iterating over all the pages in a
+ * given pair. This value can be retrieved by MTD users using the
+ * mtd_pairing_groups() helper.
+ *
+ * Examples are given in the mtd_pairing_info_to_wunit() and
+ * mtd_wunit_to_pairing_info() documentation.
+ */
+struct mtd_pairing_scheme {
+	int ngroups;
+	int (*get_info)(struct mtd_info *mtd, int wunit,
+			struct mtd_pairing_info *info);
+	int (*get_wunit)(struct mtd_info *mtd,
+			 const struct mtd_pairing_info *info);
+};
+
 /*
  * Internal ECC layout control structure. For historical reasons, there is a
  * similar, smaller struct nand_ecclayout_user (in mtd-abi.h) that is retained
@@ -148,6 +224,43 @@ struct nand_ecclayout {
 
 struct module;	/* only needed for owner field in mtd_info */
 
+/**
+ * struct mtd_part - MTD partition specific fields
+ *
+ * @node: list node used to add an MTD partition to the parent partition list
+ * @offset: offset of the partition relatively to the parent offset
+ * @size: partition size. Should be equal to mtd->size unless
+ *	  MTD_SLC_ON_MLC_EMULATION is set
+ * @flags: original flags (before the mtdpart logic decided to tweak them based
+ *	   on flash constraints, like eraseblock/pagesize alignment)
+ *
+ * This struct is embedded in mtd_info and contains partition-specific
+ * properties/fields.
+ */
+struct mtd_part {
+	struct list_head node;
+	u64 offset;
+	u64 size;
+	u32 flags;
+};
+
+/**
+ * struct mtd_master - MTD master specific fields
+ *
+ * @partitions_lock: lock protecting accesses to the partition list. Protects
+ *		     not only the master partition list, but also all
+ *		     sub-partitions.
+ * @suspended: et to 1 when the device is suspended, 0 otherwise
+ *
+ * This struct is embedded in mtd_info and contains master-specific
+ * properties/fields. The master is the root MTD device from the MTD partition
+ * point of view.
+ */
+struct mtd_master {
+	struct mutex partitions_lock;
+	unsigned int suspended : 1;
+};
+
 struct mtd_info {
 	u_char type;
 	uint32_t flags;
@@ -211,6 +324,9 @@ struct mtd_info {
 	/* OOB layout description */
 	const struct mtd_ooblayout_ops *ooblayout;
 
+	/* NAND pairing scheme, only provided for MLC/TLC NANDs */
+	const struct mtd_pairing_scheme *pairing;
+
 	/* ECC layout structure pointer - read only! */
 	struct nand_ecclayout *ecclayout;
 
@@ -329,8 +445,31 @@ struct mtd_info {
 	 * MTD device can itself be a partition).
 	 */
 	struct list_head partitions;
+
+	union {
+		struct mtd_part part;
+		struct mtd_master master;
+	};
 };
 
+static inline struct mtd_info *mtd_get_master(struct mtd_info *mtd)
+{
+	while (mtd->parent)
+		mtd = mtd->parent;
+
+	return mtd;
+}
+
+static inline u64 mtd_get_master_ofs(struct mtd_info *mtd, u64 ofs)
+{
+	while (mtd->parent) {
+		ofs += mtd->offset;
+		mtd = mtd->parent;
+	}
+
+	return ofs;
+}
+
 #if IS_ENABLED(CONFIG_DM)
 static inline void mtd_set_ofnode(struct mtd_info *mtd, ofnode node)
 {
@@ -391,11 +530,22 @@ static inline void mtd_set_ooblayout(struct mtd_info *mtd,
 	mtd->ooblayout = ooblayout;
 }
 
+static inline void mtd_set_pairing_scheme(struct mtd_info *mtd,
+				const struct mtd_pairing_scheme *pairing)
+{
+	mtd->pairing = pairing;
+}
+
 static inline u32 mtd_oobavail(struct mtd_info *mtd, struct mtd_oob_ops *ops)
 {
 	return ops->mode == MTD_OPS_AUTO_OOB ? mtd->oobavail : mtd->oobsize;
 }
 
+int mtd_wunit_to_pairing_info(struct mtd_info *mtd, int wunit,
+			      struct mtd_pairing_info *info);
+int mtd_pairing_info_to_wunit(struct mtd_info *mtd,
+			      const struct mtd_pairing_info *info);
+int mtd_pairing_groups(struct mtd_info *mtd);
 int mtd_erase(struct mtd_info *mtd, struct erase_info *instr);
 #ifndef __UBOOT__
 int mtd_point(struct mtd_info *mtd, loff_t from, size_t len, size_t *retlen,
@@ -433,8 +583,10 @@ int mtd_writev(struct mtd_info *mtd, const struct kvec *vecs,
 
 static inline void mtd_sync(struct mtd_info *mtd)
 {
-	if (mtd->_sync)
-		mtd->_sync(mtd);
+	struct mtd_info *master = mtd_get_master(mtd);
+
+	if (master->_sync)
+		master->_sync(master);
 }
 
 int mtd_lock(struct mtd_info *mtd, loff_t ofs, uint64_t len);
@@ -447,13 +599,15 @@ int mtd_block_markbad(struct mtd_info *mtd, loff_t ofs);
 #ifndef __UBOOT__
 static inline int mtd_suspend(struct mtd_info *mtd)
 {
-	return mtd->_suspend ? mtd->_suspend(mtd) : 0;
+	struct mtd_info *master = mtd_get_master(mtd);
+	return master->_suspend ? master->_suspend(master) : 0;
 }
 
 static inline void mtd_resume(struct mtd_info *mtd)
 {
-	if (mtd->_resume)
-		mtd->_resume(mtd);
+	struct mtd_info *master = mtd_get_master(mtd);
+	if (master->_resume)
+		master->_resume(master);
 }
 #endif
 
@@ -487,9 +641,29 @@ static inline uint32_t mtd_mod_by_ws(uint64_t sz, struct mtd_info *mtd)
 	return do_div(sz, mtd->writesize);
 }
 
+static inline int mtd_wunit_per_eb(struct mtd_info *mtd)
+{
+	struct mtd_info *master = mtd_get_master(mtd);
+
+	return master->erasesize / mtd->writesize;
+}
+
+static inline int mtd_offset_to_wunit(struct mtd_info *mtd, loff_t offs)
+{
+	return mtd_div_by_ws(mtd_mod_by_eb(offs, mtd), mtd);
+}
+
+static inline loff_t mtd_wunit_to_offset(struct mtd_info *mtd, loff_t base,
+					 int wunit)
+{
+	return base + (wunit * mtd->writesize);
+}
+
 static inline int mtd_has_oob(const struct mtd_info *mtd)
 {
-	return mtd->_read_oob && mtd->_write_oob;
+	struct mtd_info *master = mtd_get_master((struct mtd_info *)mtd);
+
+	return master->_read_oob && master->_write_oob;
 }
 
 static inline int mtd_type_is_nand(const struct mtd_info *mtd)
@@ -499,7 +673,9 @@ static inline int mtd_type_is_nand(const struct mtd_info *mtd)
 
 static inline int mtd_can_have_bb(const struct mtd_info *mtd)
 {
-	return !!mtd->_block_isbad;
+	struct mtd_info *master = mtd_get_master((struct mtd_info *)mtd);
+
+	return !!master->_block_isbad;
 }
 
 	/* Kernel-side ioctl definitions */
diff --git a/include/linux/mtd/partitions.h b/include/linux/mtd/partitions.h
index 3822237f2a..34eec5f67a 100644
--- a/include/linux/mtd/partitions.h
+++ b/include/linux/mtd/partitions.h
@@ -20,6 +20,12 @@
  *
  * For each partition, these fields are available:
  * name: string that will be used to label the partition's MTD device.
+ * types: some partitions can be containers using specific format to describe
+ *	embedded subpartitions / volumes. E.g. many home routers use "firmware"
+ *	partition that contains at least kernel and rootfs. In such case an
+ *	extra parser is needed that will detect these dynamic partitions and
+ *	report them to the MTD subsystem. If set this property stores an array
+ *	of parser names to use when looking for subpartitions.
  * size: the partition size; if defined as MTDPART_SIZ_FULL, the partition
  * 	will extend to the end of the master MTD device.
  * offset: absolute starting position within the master MTD device; if
@@ -31,6 +37,7 @@
  * 	master MTD flag set for the corresponding MTD partition.
  * 	For example, to force a read-only partition, simply adding
  * 	MTD_WRITEABLE to the mask_flags will do the trick.
+ * add_flags: contains flags to add to the parent flags
  *
  * Note: writeable partitions require their size and offset be
  * erasesize aligned (e.g. use MTDPART_OFS_NEXTBLK).
@@ -38,9 +45,11 @@
 
 struct mtd_partition {
 	const char *name;		/* identifier string */
+	const char *const *types;	/* names of parsers to use if any */
 	uint64_t size;			/* partition size */
 	uint64_t offset;		/* offset within the master MTD space */
 	uint32_t mask_flags;		/* master MTD flags to mask out for this partition */
+	uint32_t add_flags;		/* flags to add to the partition */
 	struct nand_ecclayout *ecclayout;	/* out of band layout for this partition (NAND only) */
 };
 
@@ -53,7 +62,6 @@ struct mtd_partition {
 struct mtd_info;
 struct device_node;
 
-#ifndef __UBOOT__
 /**
  * struct mtd_part_parser_data - used to pass data to MTD partition parsers.
  * @origin: for RedBoot, start address of MTD device
@@ -64,7 +72,6 @@ struct mtd_part_parser_data {
 	struct device_node *of_node;
 };
 
-
 /*
  * Functions dealing with the various ways of partitioning the space
  */
@@ -73,13 +80,22 @@ struct mtd_part_parser {
 	struct list_head list;
 	struct module *owner;
 	const char *name;
-	int (*parse_fn)(struct mtd_info *, struct mtd_partition **,
+	const struct udevice_id *of_match_table;
+	int (*parse_fn)(struct mtd_info *, const struct mtd_partition **,
 			struct mtd_part_parser_data *);
+	void (*cleanup)(const struct mtd_partition *pparts, int nr_parts);
+};
+
+/* Container for passing around a set of parsed partitions */
+struct mtd_partitions {
+	const struct mtd_partition *parts;
+	int nr_parts;
+	const struct mtd_part_parser *parser;
 };
 
+
 extern void register_mtd_parser(struct mtd_part_parser *parser);
 extern void deregister_mtd_parser(struct mtd_part_parser *parser);
-#endif
 
 int mtd_add_partition(struct mtd_info *master, const char *name,
 		      long long offset, long long length);
diff --git a/include/mtd/mtd-abi.h b/include/mtd/mtd-abi.h
index 8d9761050f..2f6ce18795 100644
--- a/include/mtd/mtd-abi.h
+++ b/include/mtd/mtd-abi.h
@@ -94,6 +94,7 @@ struct mtd_write_req {
 #define MTD_BIT_WRITEABLE	0x800	/* Single bits can be flipped */
 #define MTD_NO_ERASE		0x1000	/* No erase necessary */
 #define MTD_POWERUP_LOCK	0x2000	/* Always locked after reset */
+#define MTD_SLC_ON_MLC_EMULATION 0x4000	/* Emulate SLC behavior on MLC NANDs */
 
 /* Some common devices / combinations of capabilities */
 #define MTD_CAP_ROM		0
-- 
2.30.2

[RFCv3 2/3] mtd: Add pairing info for Toshiba TC58TEG5DCLTA00 NAND

[Chris Morgan]

From: Chris Morgan <macromorgan@hotmail.com>

Add pairing info for the Toshiba TC58TEG5DCLTA00 NAND so that we may
utilize slc-mode emulation for enhanced reliability and compatibility
with upstream Linux.

Signed-off-by: Chris Morgan <macromorgan@hotmail.com>
---
 drivers/mtd/nand/raw/nand_base.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/mtd/nand/raw/nand_base.c b/drivers/mtd/nand/raw/nand_base.c
index 9e50dce40b..70b4a76bad 100644
--- a/drivers/mtd/nand/raw/nand_base.c
+++ b/drivers/mtd/nand/raw/nand_base.c
@@ -4546,6 +4546,15 @@ ident_done:
 			break;
 	}
 
+	/* Enable slc-mode on TC58TEG5DCLTA00 to enable upstream Linux */
+	if (id_data[0] == NAND_MFR_TOSHIBA && id_data[1] == 0xd7
+	    && id_data[2] == 0x84 && id_data[3] == 0x93
+	    && id_data[4] == 0x72 && id_data[5] == 0x51
+	    && id_data[6] == 0x08 && id_data[7] == 0x04) {
+		chip->options |= NAND_NEED_SCRAMBLING;
+		mtd_set_pairing_scheme(mtd, &dist3_pairing_scheme);
+	}
+
 	if (chip->options & NAND_BUSWIDTH_AUTO) {
 		WARN_ON(chip->options & NAND_BUSWIDTH_16);
 		chip->options |= busw;
-- 
2.30.2

[RFCv3 3/3] mtd: Add pairing info for Hynix H27UCG8T2ETR NAND

[Chris Morgan]

From: Chris Morgan <macromorgan@hotmail.com>

Add pairing info for the Hynix H27UCG8T2ETR NAND so that we may utilize
slc-mode emulation for enhanced reliability and compatibility with
upstream Linux.

I have checked with the team at #mtd and they can confirm based on
a cursory reading of the datasheet for this flash chip it appears to
use the same dist3 pairing scheme used by the Toshiba NAND which
already supports dist3 today in mainline Linux.

Signed-off-by: Chris Morgan <macromorgan@hotmail.com>
---
 drivers/mtd/nand/raw/nand_base.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/drivers/mtd/nand/raw/nand_base.c b/drivers/mtd/nand/raw/nand_base.c
index 70b4a76bad..fc6ee36514 100644
--- a/drivers/mtd/nand/raw/nand_base.c
+++ b/drivers/mtd/nand/raw/nand_base.c
@@ -4555,6 +4555,14 @@ ident_done:
 		mtd_set_pairing_scheme(mtd, &dist3_pairing_scheme);
 	}
 
+	/* Enable slc-mode on H27UCG8T2ETR to enable upstream Linux */
+	if (id_data[0] == NAND_MFR_HYNIX && id_data[1] == 0xde
+	    && id_data[2] == 0x14 && id_data[3] == 0xa7
+	    && id_data[4] == 0x42 && id_data[5] == 0x4a) {
+		chip->options |= NAND_NEED_SCRAMBLING;
+		mtd_set_pairing_scheme(mtd, &dist3_pairing_scheme);
+	}
+
 	if (chip->options & NAND_BUSWIDTH_AUTO) {
 		WARN_ON(chip->options & NAND_BUSWIDTH_16);
 		chip->options |= busw;
-- 
2.30.2