Merge tag 'mtd/spi-mem-ecc-for-5.18' into mtd/next

Topic branch bringing-in changes related to the support of ECC engines
that can be used by SPI controllers to manage SPI NANDs as well as
possibly by parallel NAND controllers. In particular, it brings support
for Macronix ECC engine that can be used with Macronix SPI controller.

The changes touch the NAND core, the NAND ECC core, the spi-mem layer, a
SPI controller driver and add a new NAND ECC driver, as well as a number
of binding updates.

Binding changes:
* Vendor prefixes: Clarify Macronix prefix
* SPI NAND: Convert spi-nand description file to yaml
* Raw NAND chip: Create a NAND chip description
* Raw NAND controller:
  - Harmonize the property types
  - Fix a comment in the examples
  - Fix the reg property description
* Describe Macronix NAND ECC engine
* Macronix SPI controller:
  - Document the nand-ecc-engine property
  - Convert to yaml
  - The interrupt property is not mandatory

NAND core changes:
* ECC:
  - Add infrastructure to support hardware engines
  - Add a new helper to retrieve the ECC context
  - Provide a helper to retrieve a pilelined engine device

NAND-ECC changes:
* Macronix ECC engine:
  - Add Macronix external ECC engine support
  - Support SPI pipelined mode

SPI-NAND core changes:
* Delay a little bit the dirmap creation
* Create direct mapping descriptors for ECC operations

SPI-NAND driver changes:
* macronix: Use random program load

SPI changes:
* Macronix SPI controller:
  - Fix the transmit path
  - Create a helper to configure the controller before an operation
  - Create a helper to ease the start of an operation
  - Add support for direct mapping
  - Add support for pipelined ECC operations
* spi-mem:
  - Introduce a capability structure
  - Check the controller extra capabilities
  - cadence-quadspi/mxic: Provide capability structures
  - Kill the spi_mem_dtr_supports_op() helper
  - Add an ecc parameter to the spi_mem_op structure

11 files changed, 1359 insertions, 92 deletions

diff --git a/drivers/mtd/nand/Kconfig b/drivers/mtd/nand/Kconfig
index b40455234cbd..8431292ff49d 100644
--- a/drivers/mtd/nand/Kconfig
+++ b/drivers/mtd/nand/Kconfig
@@ -46,6 +46,12 @@ config MTD_NAND_ECC_SW_BCH
 	  ECC codes. They are used with NAND devices requiring more than 1 bit
 	  of error correction.
 
+config MTD_NAND_ECC_MXIC
+	bool "Macronix external hardware ECC engine"
+	select MTD_NAND_ECC
+	help
+	  This enables support for the hardware ECC engine from Macronix.
+
 endmenu
 
 endmenu
diff --git a/drivers/mtd/nand/Makefile b/drivers/mtd/nand/Makefile
index 1c0b46960eb1..a4e6b7ae0614 100644
--- a/drivers/mtd/nand/Makefile
+++ b/drivers/mtd/nand/Makefile
@@ -10,3 +10,4 @@ obj-y	+= spi/
 nandcore-$(CONFIG_MTD_NAND_ECC) += ecc.o
 nandcore-$(CONFIG_MTD_NAND_ECC_SW_HAMMING) += ecc-sw-hamming.o
 nandcore-$(CONFIG_MTD_NAND_ECC_SW_BCH) += ecc-sw-bch.o
+nandcore-$(CONFIG_MTD_NAND_ECC_MXIC) += ecc-mxic.o
diff --git a/drivers/mtd/nand/core.c b/drivers/mtd/nand/core.c
index 3ff99a42a63f..dbd7b06524b3 100644
--- a/drivers/mtd/nand/core.c
+++ b/drivers/mtd/nand/core.c
@@ -235,7 +235,9 @@ static int nanddev_get_ecc_engine(struct nand_device *nand)
 		nand->ecc.engine = nand_ecc_get_on_die_hw_engine(nand);
 		break;
 	case NAND_ECC_ENGINE_TYPE_ON_HOST:
-		pr_err("On-host hardware ECC engines not supported yet\n");
+		nand->ecc.engine = nand_ecc_get_on_host_hw_engine(nand);
+		if (PTR_ERR(nand->ecc.engine) == -EPROBE_DEFER)
+			return -EPROBE_DEFER;
 		break;
 	default:
 		pr_err("Missing ECC engine type\n");
@@ -255,7 +257,7 @@ static int nanddev_put_ecc_engine(struct nand_device *nand)
 {
 	switch (nand->ecc.ctx.conf.engine_type) {
 	case NAND_ECC_ENGINE_TYPE_ON_HOST:
-		pr_err("On-host hardware ECC engines not supported yet\n");
+		nand_ecc_put_on_host_hw_engine(nand);
 		break;
 	case NAND_ECC_ENGINE_TYPE_NONE:
 	case NAND_ECC_ENGINE_TYPE_SOFT:
@@ -300,7 +302,9 @@ int nanddev_ecc_engine_init(struct nand_device *nand)
 	/* Look for the ECC engine to use */
 	ret = nanddev_get_ecc_engine(nand);
 	if (ret) {
-		pr_err("No ECC engine found\n");
+		if (ret != -EPROBE_DEFER)
+			pr_err("No ECC engine found\n");
+
 		return ret;
 	}
 
diff --git a/drivers/mtd/nand/ecc-mxic.c b/drivers/mtd/nand/ecc-mxic.c
new file mode 100644
index 000000000000..c122139255e5
--- /dev/null
+++ b/drivers/mtd/nand/ecc-mxic.c
@@ -0,0 +1,879 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Support for Macronix external hardware ECC engine for NAND devices, also
+ * called DPE for Data Processing Engine.
+ *
+ * Copyright © 2019 Macronix
+ * Author: Miquel Raynal <miquel.raynal@bootlin.com>
+ */
+
+#include <linux/dma-mapping.h>
+#include <linux/init.h>
+#include <linux/interrupt.h>
+#include <linux/io.h>
+#include <linux/iopoll.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/mtd/mtd.h>
+#include <linux/mtd/nand.h>
+#include <linux/mtd/nand-ecc-mxic.h>
+#include <linux/mutex.h>
+#include <linux/of_device.h>
+#include <linux/of_platform.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+
+/* DPE Configuration */
+#define DP_CONFIG 0x00
+#define   ECC_EN BIT(0)
+#define   ECC_TYP(idx) (((idx) << 3) & GENMASK(6, 3))
+/* DPE Interrupt Status */
+#define INTRPT_STS 0x04
+#define   TRANS_CMPLT BIT(0)
+#define   SDMA_MAIN BIT(1)
+#define   SDMA_SPARE BIT(2)
+#define   ECC_ERR BIT(3)
+#define   TO_SPARE BIT(4)
+#define   TO_MAIN BIT(5)
+/* DPE Interrupt Status Enable */
+#define INTRPT_STS_EN 0x08
+/* DPE Interrupt Signal Enable */
+#define INTRPT_SIG_EN 0x0C
+/* Host Controller Configuration */
+#define HC_CONFIG 0x10
+#define   DEV2MEM 0 /* TRANS_TYP_DMA in the spec */
+#define   MEM2MEM BIT(4) /* TRANS_TYP_IO in the spec */
+#define   MAPPING BIT(5) /* TRANS_TYP_MAPPING in the spec */
+#define   ECC_PACKED 0 /* LAYOUT_TYP_INTEGRATED in the spec */
+#define   ECC_INTERLEAVED BIT(2) /* LAYOUT_TYP_DISTRIBUTED in the spec */
+#define   BURST_TYP_FIXED 0
+#define   BURST_TYP_INCREASING BIT(0)
+/* Host Controller Slave Address */
+#define HC_SLV_ADDR 0x14
+/* ECC Chunk Size */
+#define CHUNK_SIZE 0x20
+/* Main Data Size */
+#define MAIN_SIZE 0x24
+/* Spare Data Size */
+#define SPARE_SIZE 0x28
+#define   META_SZ(reg) ((reg) & GENMASK(7, 0))
+#define   PARITY_SZ(reg) (((reg) & GENMASK(15, 8)) >> 8)
+#define   RSV_SZ(reg) (((reg) & GENMASK(23, 16)) >> 16)
+#define   SPARE_SZ(reg) ((reg) >> 24)
+/* ECC Chunk Count */
+#define CHUNK_CNT 0x30
+/* SDMA Control */
+#define SDMA_CTRL 0x40
+#define   WRITE_NAND 0
+#define   READ_NAND BIT(1)
+#define   CONT_NAND BIT(29)
+#define   CONT_SYSM BIT(30) /* Continue System Memory? */
+#define   SDMA_STRT BIT(31)
+/* SDMA Address of Main Data */
+#define SDMA_MAIN_ADDR 0x44
+/* SDMA Address of Spare Data */
+#define SDMA_SPARE_ADDR 0x48
+/* DPE Version Number */
+#define DP_VER 0xD0
+#define   DP_VER_OFFSET 16
+
+/* Status bytes between each chunk of spare data */
+#define STAT_BYTES 4
+#define   NO_ERR 0x00
+#define   MAX_CORR_ERR 0x28
+#define   UNCORR_ERR 0xFE
+#define   ERASED_CHUNK 0xFF
+
+struct mxic_ecc_engine {
+	struct device *dev;
+	void __iomem *regs;
+	int irq;
+	struct completion complete;
+	struct nand_ecc_engine external_engine;
+	struct nand_ecc_engine pipelined_engine;
+	struct mutex lock;
+};
+
+struct mxic_ecc_ctx {
+	/* ECC machinery */
+	unsigned int data_step_sz;
+	unsigned int oob_step_sz;
+	unsigned int parity_sz;
+	unsigned int meta_sz;
+	u8 *status;
+	int steps;
+
+	/* DMA boilerplate */
+	struct nand_ecc_req_tweak_ctx req_ctx;
+	u8 *oobwithstat;
+	struct scatterlist sg[2];
+	struct nand_page_io_req *req;
+	unsigned int pageoffs;
+};
+
+static struct mxic_ecc_engine *ext_ecc_eng_to_mxic(struct nand_ecc_engine *eng)
+{
+	return container_of(eng, struct mxic_ecc_engine, external_engine);
+}
+
+static struct mxic_ecc_engine *pip_ecc_eng_to_mxic(struct nand_ecc_engine *eng)
+{
+	return container_of(eng, struct mxic_ecc_engine, pipelined_engine);
+}
+
+static struct mxic_ecc_engine *nand_to_mxic(struct nand_device *nand)
+{
+	struct nand_ecc_engine *eng = nand->ecc.engine;
+
+	if (eng->integration == NAND_ECC_ENGINE_INTEGRATION_EXTERNAL)
+		return ext_ecc_eng_to_mxic(eng);
+	else
+		return pip_ecc_eng_to_mxic(eng);
+}
+
+static int mxic_ecc_ooblayout_ecc(struct mtd_info *mtd, int section,
+				  struct mtd_oob_region *oobregion)
+{
+	struct nand_device *nand = mtd_to_nanddev(mtd);
+	struct mxic_ecc_ctx *ctx = nand_to_ecc_ctx(nand);
+
+	if (section < 0 || section >= ctx->steps)
+		return -ERANGE;
+
+	oobregion->offset = (section * ctx->oob_step_sz) + ctx->meta_sz;
+	oobregion->length = ctx->parity_sz;
+
+	return 0;
+}
+
+static int mxic_ecc_ooblayout_free(struct mtd_info *mtd, int section,
+				   struct mtd_oob_region *oobregion)
+{
+	struct nand_device *nand = mtd_to_nanddev(mtd);
+	struct mxic_ecc_ctx *ctx = nand_to_ecc_ctx(nand);
+
+	if (section < 0 || section >= ctx->steps)
+		return -ERANGE;
+
+	if (!section) {
+		oobregion->offset = 2;
+		oobregion->length = ctx->meta_sz - 2;
+	} else {
+		oobregion->offset = section * ctx->oob_step_sz;
+		oobregion->length = ctx->meta_sz;
+	}
+
+	return 0;
+}
+
+static const struct mtd_ooblayout_ops mxic_ecc_ooblayout_ops = {
+	.ecc = mxic_ecc_ooblayout_ecc,
+	.free = mxic_ecc_ooblayout_free,
+};
+
+static void mxic_ecc_disable_engine(struct mxic_ecc_engine *mxic)
+{
+	u32 reg;
+
+	reg = readl(mxic->regs + DP_CONFIG);
+	reg &= ~ECC_EN;
+	writel(reg, mxic->regs + DP_CONFIG);
+}
+
+static void mxic_ecc_enable_engine(struct mxic_ecc_engine *mxic)
+{
+	u32 reg;
+
+	reg = readl(mxic->regs + DP_CONFIG);
+	reg |= ECC_EN;
+	writel(reg, mxic->regs + DP_CONFIG);
+}
+
+static void mxic_ecc_disable_int(struct mxic_ecc_engine *mxic)
+{
+	writel(0, mxic->regs + INTRPT_SIG_EN);
+}
+
+static void mxic_ecc_enable_int(struct mxic_ecc_engine *mxic)
+{
+	writel(TRANS_CMPLT, mxic->regs + INTRPT_SIG_EN);
+}
+
+static irqreturn_t mxic_ecc_isr(int irq, void *dev_id)
+{
+	struct mxic_ecc_engine *mxic = dev_id;
+	u32 sts;
+
+	sts = readl(mxic->regs + INTRPT_STS);
+	if (!sts)
+		return IRQ_NONE;
+
+	if (sts & TRANS_CMPLT)
+		complete(&mxic->complete);
+
+	writel(sts, mxic->regs + INTRPT_STS);
+
+	return IRQ_HANDLED;
+}
+
+static int mxic_ecc_init_ctx(struct nand_device *nand, struct device *dev)
+{
+	struct mxic_ecc_engine *mxic = nand_to_mxic(nand);
+	struct nand_ecc_props *conf = &nand->ecc.ctx.conf;
+	struct nand_ecc_props *reqs = &nand->ecc.requirements;
+	struct nand_ecc_props *user = &nand->ecc.user_conf;
+	struct mtd_info *mtd = nanddev_to_mtd(nand);
+	int step_size = 0, strength = 0, desired_correction = 0, steps, idx;
+	int possible_strength[] = {4, 8, 40, 48};
+	int spare_size[] = {32, 32, 96, 96};
+	struct mxic_ecc_ctx *ctx;
+	u32 spare_reg;
+	int ret;
+
+	ctx = devm_kzalloc(dev, sizeof(*ctx), GFP_KERNEL);
+	if (!ctx)
+		return -ENOMEM;
+
+	nand->ecc.ctx.priv = ctx;
+
+	/* Only large page NAND chips may use BCH */
+	if (mtd->oobsize < 64) {
+		pr_err("BCH cannot be used with small page NAND chips\n");
+		return -EINVAL;
+	}
+
+	mtd_set_ooblayout(mtd, &mxic_ecc_ooblayout_ops);
+
+	/* Enable all status bits */
+	writel(TRANS_CMPLT | SDMA_MAIN | SDMA_SPARE | ECC_ERR |
+	       TO_SPARE | TO_MAIN, mxic->regs + INTRPT_STS_EN);
+
+	/* Configure the correction depending on the NAND device topology */
+	if (user->step_size && user->strength) {
+		step_size = user->step_size;
+		strength = user->strength;
+	} else if (reqs->step_size && reqs->strength) {
+		step_size = reqs->step_size;
+		strength = reqs->strength;
+	}
+
+	if (step_size && strength) {
+		steps = mtd->writesize / step_size;
+		desired_correction = steps * strength;
+	}
+
+	/* Step size is fixed to 1kiB, strength may vary (4 possible values) */
+	conf->step_size = SZ_1K;
+	steps = mtd->writesize / conf->step_size;
+
+	ctx->status = devm_kzalloc(dev, steps * sizeof(u8), GFP_KERNEL);
+	if (!ctx->status)
+		return -ENOMEM;
+
+	if (desired_correction) {
+		strength = desired_correction / steps;
+
+		for (idx = 0; idx < ARRAY_SIZE(possible_strength); idx++)
+			if (possible_strength[idx] >= strength)
+				break;
+
+		idx = min_t(unsigned int, idx,
+			    ARRAY_SIZE(possible_strength) - 1);
+	} else {
+		/* Missing data, maximize the correction */
+		idx = ARRAY_SIZE(possible_strength) - 1;
+	}
+
+	/* Tune the selected strength until it fits in the OOB area */
+	for (; idx >= 0; idx--) {
+		if (spare_size[idx] * steps <= mtd->oobsize)
+			break;
+	}
+
+	/* This engine cannot be used with this NAND device */
+	if (idx < 0)
+		return -EINVAL;
+
+	/* Configure the engine for the desired strength */
+	writel(ECC_TYP(idx), mxic->regs + DP_CONFIG);
+	conf->strength = possible_strength[idx];
+	spare_reg = readl(mxic->regs + SPARE_SIZE);
+
+	ctx->steps = steps;
+	ctx->data_step_sz = mtd->writesize / steps;
+	ctx->oob_step_sz = mtd->oobsize / steps;
+	ctx->parity_sz = PARITY_SZ(spare_reg);
+	ctx->meta_sz = META_SZ(spare_reg);
+
+	/* Ensure buffers will contain enough bytes to store the STAT_BYTES */
+	ctx->req_ctx.oob_buffer_size = nanddev_per_page_oobsize(nand) +
+					(ctx->steps * STAT_BYTES);
+	ret = nand_ecc_init_req_tweaking(&ctx->req_ctx, nand);
+	if (ret)
+		return ret;
+
+	ctx->oobwithstat = kmalloc(mtd->oobsize + (ctx->steps * STAT_BYTES),
+				   GFP_KERNEL);
+	if (!ctx->oobwithstat) {
+		ret = -ENOMEM;
+		goto cleanup_req_tweak;
+	}
+
+	sg_init_table(ctx->sg, 2);
+
+	/* Configuration dump and sanity checks */
+	dev_err(dev, "DPE version number: %d\n",
+		readl(mxic->regs + DP_VER) >> DP_VER_OFFSET);
+	dev_err(dev, "Chunk size: %d\n", readl(mxic->regs + CHUNK_SIZE));
+	dev_err(dev, "Main size: %d\n", readl(mxic->regs + MAIN_SIZE));
+	dev_err(dev, "Spare size: %d\n", SPARE_SZ(spare_reg));
+	dev_err(dev, "Rsv size: %ld\n", RSV_SZ(spare_reg));
+	dev_err(dev, "Parity size: %d\n", ctx->parity_sz);
+	dev_err(dev, "Meta size: %d\n", ctx->meta_sz);
+
+	if ((ctx->meta_sz + ctx->parity_sz + RSV_SZ(spare_reg)) !=
+	    SPARE_SZ(spare_reg)) {
+		dev_err(dev, "Wrong OOB configuration: %d + %d + %ld != %d\n",
+			ctx->meta_sz, ctx->parity_sz, RSV_SZ(spare_reg),
+			SPARE_SZ(spare_reg));
+		ret = -EINVAL;
+		goto free_oobwithstat;
+	}
+
+	if (ctx->oob_step_sz != SPARE_SZ(spare_reg)) {
+		dev_err(dev, "Wrong OOB configuration: %d != %d\n",
+			ctx->oob_step_sz, SPARE_SZ(spare_reg));
+		ret = -EINVAL;
+		goto free_oobwithstat;
+	}
+
+	return 0;
+
+free_oobwithstat:
+	kfree(ctx->oobwithstat);
+cleanup_req_tweak:
+	nand_ecc_cleanup_req_tweaking(&ctx->req_ctx);
+
+	return ret;
+}
+
+static int mxic_ecc_init_ctx_external(struct nand_device *nand)
+{
+	struct mxic_ecc_engine *mxic = nand_to_mxic(nand);
+	struct device *dev = nand->ecc.engine->dev;
+	int ret;
+
+	dev_info(dev, "Macronix ECC engine in external mode\n");
+
+	ret = mxic_ecc_init_ctx(nand, dev);
+	if (ret)
+		return ret;
+
+	/* Trigger each step manually */
+	writel(1, mxic->regs + CHUNK_CNT);
+	writel(BURST_TYP_INCREASING | ECC_PACKED | MEM2MEM,
+	       mxic->regs + HC_CONFIG);
+
+	return 0;
+}
+
+static int mxic_ecc_init_ctx_pipelined(struct nand_device *nand)
+{
+	struct mxic_ecc_engine *mxic = nand_to_mxic(nand);
+	struct mxic_ecc_ctx *ctx;
+	struct device *dev;
+	int ret;
+
+	dev = nand_ecc_get_engine_dev(nand->ecc.engine->dev);
+	if (!dev)
+		return -EINVAL;
+
+	dev_info(dev, "Macronix ECC engine in pipelined/mapping mode\n");
+
+	ret = mxic_ecc_init_ctx(nand, dev);
+	if (ret)
+		return ret;
+
+	ctx = nand_to_ecc_ctx(nand);
+
+	/* All steps should be handled in one go directly by the internal DMA */
+	writel(ctx->steps, mxic->regs + CHUNK_CNT);
+
+	/*
+	 * Interleaved ECC scheme cannot be used otherwise factory bad block
+	 * markers would be lost. A packed layout is mandatory.
+	 */
+	writel(BURST_TYP_INCREASING | ECC_PACKED | MAPPING,
+	       mxic->regs + HC_CONFIG);
+
+	return 0;
+}
+
+static void mxic_ecc_cleanup_ctx(struct nand_device *nand)
+{
+	struct mxic_ecc_ctx *ctx = nand_to_ecc_ctx(nand);
+
+	if (ctx) {
+		nand_ecc_cleanup_req_tweaking(&ctx->req_ctx);
+		kfree(ctx->oobwithstat);
+	}
+}
+
+static int mxic_ecc_data_xfer_wait_for_completion(struct mxic_ecc_engine *mxic)
+{
+	u32 val;
+	int ret;
+
+	if (mxic->irq) {
+		reinit_completion(&mxic->complete);
+		mxic_ecc_enable_int(mxic);
+		ret = wait_for_completion_timeout(&mxic->complete,
+						  msecs_to_jiffies(1000));
+		mxic_ecc_disable_int(mxic);
+	} else {
+		ret = readl_poll_timeout(mxic->regs + INTRPT_STS, val,
+					 val & TRANS_CMPLT, 10, USEC_PER_SEC);
+		writel(val, mxic->regs + INTRPT_STS);
+	}
+
+	if (ret) {
+		dev_err(mxic->dev, "Timeout on data xfer completion\n");
+		return -ETIMEDOUT;
+	}
+
+	return 0;
+}
+
+static int mxic_ecc_process_data(struct mxic_ecc_engine *mxic,
+				 unsigned int direction)
+{
+	unsigned int dir = (direction == NAND_PAGE_READ) ?
+			   READ_NAND : WRITE_NAND;
+	int ret;
+
+	mxic_ecc_enable_engine(mxic);
+
+	/* Trigger processing */
+	writel(SDMA_STRT | dir, mxic->regs + SDMA_CTRL);
+
+	/* Wait for completion */
+	ret = mxic_ecc_data_xfer_wait_for_completion(mxic);
+
+	mxic_ecc_disable_engine(mxic);
+
+	return ret;
+}
+
+int mxic_ecc_process_data_pipelined(struct nand_ecc_engine *eng,
+				    unsigned int direction, dma_addr_t dirmap)
+{
+	struct mxic_ecc_engine *mxic = pip_ecc_eng_to_mxic(eng);
+
+	if (dirmap)
+		writel(dirmap, mxic->regs + HC_SLV_ADDR);
+
+	return mxic_ecc_process_data(mxic, direction);
+}
+EXPORT_SYMBOL_GPL(mxic_ecc_process_data_pipelined);
+
+static void mxic_ecc_extract_status_bytes(struct mxic_ecc_ctx *ctx)
+{
+	u8 *buf = ctx->oobwithstat;
+	int next_stat_pos;
+	int step;
+
+	/* Extract the ECC status */
+	for (step = 0; step < ctx->steps; step++) {
+		next_stat_pos = ctx->oob_step_sz +
+				((STAT_BYTES + ctx->oob_step_sz) * step);
+
+		ctx->status[step] = buf[next_stat_pos];
+	}
+}
+
+static void mxic_ecc_reconstruct_oobbuf(struct mxic_ecc_ctx *ctx,
+					u8 *dst, const u8 *src)
+{
+	int step;
+
+	/* Reconstruct the OOB buffer linearly (without the ECC status bytes) */
+	for (step = 0; step < ctx->steps; step++)
+		memcpy(dst + (step * ctx->oob_step_sz),
+		       src + (step * (ctx->oob_step_sz + STAT_BYTES)),
+		       ctx->oob_step_sz);
+}
+
+static void mxic_ecc_add_room_in_oobbuf(struct mxic_ecc_ctx *ctx,
+					u8 *dst, const u8 *src)
+{
+	int step;
+
+	/* Add some space in the OOB buffer for the status bytes */
+	for (step = 0; step < ctx->steps; step++)
+		memcpy(dst + (step * (ctx->oob_step_sz + STAT_BYTES)),
+		       src + (step * ctx->oob_step_sz),
+		       ctx->oob_step_sz);
+}
+
+static int mxic_ecc_count_biterrs(struct mxic_ecc_engine *mxic,
+				  struct nand_device *nand)
+{
+	struct mxic_ecc_ctx *ctx = nand_to_ecc_ctx(nand);
+	struct mtd_info *mtd = nanddev_to_mtd(nand);
+	struct device *dev = mxic->dev;
+	unsigned int max_bf = 0;
+	bool failure = false;
+	int step;
+
+	for (step = 0; step < ctx->steps; step++) {
+		u8 stat = ctx->status[step];
+
+		if (stat == NO_ERR) {
+			dev_dbg(dev, "ECC step %d: no error\n", step);
+		} else if (stat == ERASED_CHUNK) {
+			dev_dbg(dev, "ECC step %d: erased\n", step);
+		} else if (stat == UNCORR_ERR || stat > MAX_CORR_ERR) {
+			dev_dbg(dev, "ECC step %d: uncorrectable\n", step);
+			mtd->ecc_stats.failed++;
+			failure = true;
+		} else {
+			dev_dbg(dev, "ECC step %d: %d bits corrected\n",
+				step, stat);
+			max_bf = max_t(unsigned int, max_bf, stat);
+			mtd->ecc_stats.corrected += stat;
+		}
+	}
+
+	return failure ? -EBADMSG : max_bf;
+}
+
+/* External ECC engine helpers */
+static int mxic_ecc_prepare_io_req_external(struct nand_device *nand,
+					    struct nand_page_io_req *req)
+{
+	struct mxic_ecc_engine *mxic = nand_to_mxic(nand);
+	struct mxic_ecc_ctx *ctx = nand_to_ecc_ctx(nand);
+	struct mtd_info *mtd = nanddev_to_mtd(nand);
+	int offset, nents, step, ret;
+
+	if (req->mode == MTD_OPS_RAW)
+		return 0;
+
+	nand_ecc_tweak_req(&ctx->req_ctx, req);
+	ctx->req = req;
+
+	if (req->type == NAND_PAGE_READ)
+		return 0;
+
+	mxic_ecc_add_room_in_oobbuf(ctx, ctx->oobwithstat,
+				    ctx->req->oobbuf.out);
+
+	sg_set_buf(&ctx->sg[0], req->databuf.out, req->datalen);
+	sg_set_buf(&ctx->sg[1], ctx->oobwithstat,
+		   req->ooblen + (ctx->steps * STAT_BYTES));
+
+	nents = dma_map_sg(mxic->dev, ctx->sg, 2, DMA_BIDIRECTIONAL);
+	if (!nents)
+		return -EINVAL;
+
+	mutex_lock(&mxic->lock);
+
+	for (step = 0; step < ctx->steps; step++) {
+		writel(sg_dma_address(&ctx->sg[0]) + (step * ctx->data_step_sz),
+		       mxic->regs + SDMA_MAIN_ADDR);
+		writel(sg_dma_address(&ctx->sg[1]) + (step * (ctx->oob_step_sz + STAT_BYTES)),
+		       mxic->regs + SDMA_SPARE_ADDR);
+		ret = mxic_ecc_process_data(mxic, ctx->req->type);
+		if (ret)
+			break;
+	}
+
+	mutex_unlock(&mxic->lock);
+
+	dma_unmap_sg(mxic->dev, ctx->sg, 2, DMA_BIDIRECTIONAL);
+
+	if (ret)
+		return ret;
+
+	/* Retrieve the calculated ECC bytes */
+	for (step = 0; step < ctx->steps; step++) {
+		offset = ctx->meta_sz + (step * ctx->oob_step_sz);
+		mtd_ooblayout_get_eccbytes(mtd,
+					   (u8 *)ctx->req->oobbuf.out + offset,
+					   ctx->oobwithstat + (step * STAT_BYTES),
+					   step * ctx->parity_sz,
+					   ctx->parity_sz);
+	}
+
+	return 0;
+}
+
+static int mxic_ecc_finish_io_req_external(struct nand_device *nand,
+					   struct nand_page_io_req *req)
+{
+	struct mxic_ecc_engine *mxic = nand_to_mxic(nand);
+	struct mxic_ecc_ctx *ctx = nand_to_ecc_ctx(nand);
+	int nents, step, ret;
+
+	if (req->mode == MTD_OPS_RAW)
+		return 0;
+
+	if (req->type == NAND_PAGE_WRITE) {
+		nand_ecc_restore_req(&ctx->req_ctx, req);
+		return 0;
+	}
+
+	/* Copy the OOB buffer and add room for the ECC engine status bytes */
+	mxic_ecc_add_room_in_oobbuf(ctx, ctx->oobwithstat, ctx->req->oobbuf.in);
+
+	sg_set_buf(&ctx->sg[0], req->databuf.in, req->datalen);
+	sg_set_buf(&ctx->sg[1], ctx->oobwithstat,
+		   req->ooblen + (ctx->steps * STAT_BYTES));
+	nents = dma_map_sg(mxic->dev, ctx->sg, 2, DMA_BIDIRECTIONAL);
+	if (!nents)
+		return -EINVAL;
+
+	mutex_lock(&mxic->lock);
+
+	for (step = 0; step < ctx->steps; step++) {
+		writel(sg_dma_address(&ctx->sg[0]) + (step * ctx->data_step_sz),
+		       mxic->regs + SDMA_MAIN_ADDR);
+		writel(sg_dma_address(&ctx->sg[1]) + (step * (ctx->oob_step_sz + STAT_BYTES)),
+		       mxic->regs + SDMA_SPARE_ADDR);
+		ret = mxic_ecc_process_data(mxic, ctx->req->type);
+		if (ret)
+			break;
+	}
+
+	mutex_unlock(&mxic->lock);
+
+	dma_unmap_sg(mxic->dev, ctx->sg, 2, DMA_BIDIRECTIONAL);
+
+	if (ret) {
+		nand_ecc_restore_req(&ctx->req_ctx, req);
+		return ret;
+	}
+
+	/* Extract the status bytes and reconstruct the buffer */
+	mxic_ecc_extract_status_bytes(ctx);
+	mxic_ecc_reconstruct_oobbuf(ctx, ctx->req->oobbuf.in, ctx->oobwithstat);
+
+	nand_ecc_restore_req(&ctx->req_ctx, req);
+
+	return mxic_ecc_count_biterrs(mxic, nand);
+}
+
+/* Pipelined ECC engine helpers */
+static int mxic_ecc_prepare_io_req_pipelined(struct nand_device *nand,
+					     struct nand_page_io_req *req)
+{
+	struct mxic_ecc_engine *mxic = nand_to_mxic(nand);
+	struct mxic_ecc_ctx *ctx = nand_to_ecc_ctx(nand);
+	int nents;
+
+	if (req->mode == MTD_OPS_RAW)
+		return 0;
+
+	nand_ecc_tweak_req(&ctx->req_ctx, req);
+	ctx->req = req;
+
+	/* Copy the OOB buffer and add room for the ECC engine status bytes */
+	mxic_ecc_add_room_in_oobbuf(ctx, ctx->oobwithstat, ctx->req->oobbuf.in);
+
+	sg_set_buf(&ctx->sg[0], req->databuf.in, req->datalen);
+	sg_set_buf(&ctx->sg[1], ctx->oobwithstat,
+		   req->ooblen + (ctx->steps * STAT_BYTES));
+
+	nents = dma_map_sg(mxic->dev, ctx->sg, 2, DMA_BIDIRECTIONAL);
+	if (!nents)
+		return -EINVAL;
+
+	mutex_lock(&mxic->lock);
+
+	writel(sg_dma_address(&ctx->sg[0]), mxic->regs + SDMA_MAIN_ADDR);
+	writel(sg_dma_address(&ctx->sg[1]), mxic->regs + SDMA_SPARE_ADDR);
+
+	return 0;
+}
+
+static int mxic_ecc_finish_io_req_pipelined(struct nand_device *nand,
+					    struct nand_page_io_req *req)
+{
+	struct mxic_ecc_engine *mxic = nand_to_mxic(nand);
+	struct mxic_ecc_ctx *ctx = nand_to_ecc_ctx(nand);
+	int ret = 0;
+
+	if (req->mode == MTD_OPS_RAW)
+		return 0;
+
+	mutex_unlock(&mxic->lock);
+
+	dma_unmap_sg(mxic->dev, ctx->sg, 2, DMA_BIDIRECTIONAL);
+
+	if (req->type == NAND_PAGE_READ) {
+		mxic_ecc_extract_status_bytes(ctx);
+		mxic_ecc_reconstruct_oobbuf(ctx, ctx->req->oobbuf.in,
+					    ctx->oobwithstat);
+		ret = mxic_ecc_count_biterrs(mxic, nand);
+	}
+
+	nand_ecc_restore_req(&ctx->req_ctx, req);
+
+	return ret;
+}
+
+static struct nand_ecc_engine_ops mxic_ecc_engine_external_ops = {
+	.init_ctx = mxic_ecc_init_ctx_external,
+	.cleanup_ctx = mxic_ecc_cleanup_ctx,
+	.prepare_io_req = mxic_ecc_prepare_io_req_external,
+	.finish_io_req = mxic_ecc_finish_io_req_external,
+};
+
+static struct nand_ecc_engine_ops mxic_ecc_engine_pipelined_ops = {
+	.init_ctx = mxic_ecc_init_ctx_pipelined,
+	.cleanup_ctx = mxic_ecc_cleanup_ctx,
+	.prepare_io_req = mxic_ecc_prepare_io_req_pipelined,
+	.finish_io_req = mxic_ecc_finish_io_req_pipelined,
+};
+
+struct nand_ecc_engine_ops *mxic_ecc_get_pipelined_ops(void)
+{
+	return &mxic_ecc_engine_pipelined_ops;
+}
+EXPORT_SYMBOL_GPL(mxic_ecc_get_pipelined_ops);
+
+static struct platform_device *
+mxic_ecc_get_pdev(struct platform_device *spi_pdev)
+{
+	struct platform_device *eng_pdev;
+	struct device_node *np;
+
+	/* Retrieve the nand-ecc-engine phandle */
+	np = of_parse_phandle(spi_pdev->dev.of_node, "nand-ecc-engine", 0);
+	if (!np)
+		return NULL;
+
+	/* Jump to the engine's device node */
+	eng_pdev = of_find_device_by_node(np);
+	of_node_put(np);
+
+	return eng_pdev;
+}
+
+void mxic_ecc_put_pipelined_engine(struct nand_ecc_engine *eng)
+{
+	struct mxic_ecc_engine *mxic = pip_ecc_eng_to_mxic(eng);
+
+	platform_device_put(to_platform_device(mxic->dev));
+}
+EXPORT_SYMBOL_GPL(mxic_ecc_put_pipelined_engine);
+
+struct nand_ecc_engine *
+mxic_ecc_get_pipelined_engine(struct platform_device *spi_pdev)
+{
+	struct platform_device *eng_pdev;
+	struct mxic_ecc_engine *mxic;
+
+	eng_pdev = mxic_ecc_get_pdev(spi_pdev);
+	if (!eng_pdev)
+		return ERR_PTR(-ENODEV);
+
+	mxic = platform_get_drvdata(eng_pdev);
+	if (!mxic) {
+		platform_device_put(eng_pdev);
+		return ERR_PTR(-EPROBE_DEFER);
+	}
+
+	return &mxic->pipelined_engine;
+}
+EXPORT_SYMBOL_GPL(mxic_ecc_get_pipelined_engine);
+
+/*
+ * Only the external ECC engine is exported as the pipelined is SoC specific, so
+ * it is registered directly by the drivers that wrap it.
+ */
+static int mxic_ecc_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct mxic_ecc_engine *mxic;
+	int ret;
+
+	mxic = devm_kzalloc(&pdev->dev, sizeof(*mxic), GFP_KERNEL);
+	if (!mxic)
+		return -ENOMEM;
+
+	mxic->dev = &pdev->dev;
+
+	/*
+	 * Both memory regions for the ECC engine itself and the AXI slave
+	 * address are mandatory.
+	 */
+	mxic->regs = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(mxic->regs)) {
+		dev_err(&pdev->dev, "Missing memory region\n");
+		return PTR_ERR(mxic->regs);
+	}
+
+	mxic_ecc_disable_engine(mxic);
+	mxic_ecc_disable_int(mxic);
+
+	/* IRQ is optional yet much more efficient */
+	mxic->irq = platform_get_irq_byname_optional(pdev, "ecc-engine");
+	if (mxic->irq > 0) {
+		ret = devm_request_irq(&pdev->dev, mxic->irq, mxic_ecc_isr, 0,
+				       "mxic-ecc", mxic);
+		if (ret)
+			return ret;
+	} else {
+		dev_info(dev, "Invalid or missing IRQ, fallback to polling\n");
+		mxic->irq = 0;
+	}
+
+	mutex_init(&mxic->lock);
+
+	/*
+	 * In external mode, the device is the ECC engine. In pipelined mode,
+	 * the device is the host controller. The device is used to match the
+	 * right ECC engine based on the DT properties.
+	 */
+	mxic->external_engine.dev = &pdev->dev;
+	mxic->external_engine.integration = NAND_ECC_ENGINE_INTEGRATION_EXTERNAL;
+	mxic->external_engine.ops = &mxic_ecc_engine_external_ops;
+
+	nand_ecc_register_on_host_hw_engine(&mxic->external_engine);
+
+	platform_set_drvdata(pdev, mxic);
+
+	return 0;
+}
+
+static int mxic_ecc_remove(struct platform_device *pdev)
+{
+	struct mxic_ecc_engine *mxic = platform_get_drvdata(pdev);
+
+	nand_ecc_unregister_on_host_hw_engine(&mxic->external_engine);
+
+	return 0;
+}
+
+static const struct of_device_id mxic_ecc_of_ids[] = {
+	{
+		.compatible = "mxicy,nand-ecc-engine-rev3",
+	},
+	{ /* sentinel */ },
+};
+MODULE_DEVICE_TABLE(of, mxic_ecc_of_ids);
+
+static struct platform_driver mxic_ecc_driver = {
+	.driver	= {
+		.name = "mxic-nand-ecc-engine",
+		.of_match_table = mxic_ecc_of_ids,
+	},
+	.probe = mxic_ecc_probe,
+	.remove	= mxic_ecc_remove,
+};
+module_platform_driver(mxic_ecc_driver);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Miquel Raynal <miquel.raynal@bootlin.com>");
+MODULE_DESCRIPTION("Macronix NAND hardware ECC controller");
diff --git a/drivers/mtd/nand/ecc.c b/drivers/mtd/nand/ecc.c
index 6c43dfda01d4..5250764cedee 100644
--- a/drivers/mtd/nand/ecc.c
+++ b/drivers/mtd/nand/ecc.c
@@ -96,6 +96,12 @@
 #include <linux/module.h>
 #include <linux/mtd/nand.h>
 #include <linux/slab.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+#include <linux/of_platform.h>
+
+static LIST_HEAD(on_host_hw_engines);
+static DEFINE_MUTEX(on_host_hw_engines_mutex);
 
 /**
  * nand_ecc_init_ctx - Init the ECC engine context
@@ -611,6 +617,119 @@ struct nand_ecc_engine *nand_ecc_get_on_die_hw_engine(struct nand_device *nand)
 }
 EXPORT_SYMBOL(nand_ecc_get_on_die_hw_engine);
 
+int nand_ecc_register_on_host_hw_engine(struct nand_ecc_engine *engine)
+{
+	struct nand_ecc_engine *item;
+
+	if (!engine)
+		return -EINVAL;
+
+	/* Prevent multiple registrations of one engine */
+	list_for_each_entry(item, &on_host_hw_engines, node)
+		if (item == engine)
+			return 0;
+
+	mutex_lock(&on_host_hw_engines_mutex);
+	list_add_tail(&engine->node, &on_host_hw_engines);
+	mutex_unlock(&on_host_hw_engines_mutex);
+
+	return 0;
+}
+EXPORT_SYMBOL(nand_ecc_register_on_host_hw_engine);
+
+int nand_ecc_unregister_on_host_hw_engine(struct nand_ecc_engine *engine)
+{
+	if (!engine)
+		return -EINVAL;
+
+	mutex_lock(&on_host_hw_engines_mutex);
+	list_del(&engine->node);
+	mutex_unlock(&on_host_hw_engines_mutex);
+
+	return 0;
+}
+EXPORT_SYMBOL(nand_ecc_unregister_on_host_hw_engine);
+
+static struct nand_ecc_engine *nand_ecc_match_on_host_hw_engine(struct device *dev)
+{
+	struct nand_ecc_engine *item;
+
+	list_for_each_entry(item, &on_host_hw_engines, node)
+		if (item->dev == dev)
+			return item;
+
+	return NULL;
+}
+
+struct nand_ecc_engine *nand_ecc_get_on_host_hw_engine(struct nand_device *nand)
+{
+	struct nand_ecc_engine *engine = NULL;
+	struct device *dev = &nand->mtd.dev;
+	struct platform_device *pdev;
+	struct device_node *np;
+
+	if (list_empty(&on_host_hw_engines))
+		return NULL;
+
+	/* Check for an explicit nand-ecc-engine property */
+	np = of_parse_phandle(dev->of_node, "nand-ecc-engine", 0);
+	if (np) {
+		pdev = of_find_device_by_node(np);
+		if (!pdev)
+			return ERR_PTR(-EPROBE_DEFER);
+
+		engine = nand_ecc_match_on_host_hw_engine(&pdev->dev);
+		platform_device_put(pdev);
+		of_node_put(np);
+
+		if (!engine)
+			return ERR_PTR(-EPROBE_DEFER);
+	}
+
+	if (engine)
+		get_device(engine->dev);
+
+	return engine;
+}
+EXPORT_SYMBOL(nand_ecc_get_on_host_hw_engine);
+
+void nand_ecc_put_on_host_hw_engine(struct nand_device *nand)
+{
+	put_device(nand->ecc.engine->dev);
+}
+EXPORT_SYMBOL(nand_ecc_put_on_host_hw_engine);
+
+/*
+ * In the case of a pipelined engine, the device registering the ECC
+ * engine is not necessarily the ECC engine itself but may be a host controller.
+ * It is then useful to provide a helper to retrieve the right device object
+ * which actually represents the ECC engine.
+ */
+struct device *nand_ecc_get_engine_dev(struct device *host)
+{
+	struct platform_device *ecc_pdev;
+	struct device_node *np;
+
+	/*
+	 * If the device node contains this property, it means we need to follow
+	 * it in order to get the right ECC engine device we are looking for.
+	 */
+	np = of_parse_phandle(host->of_node, "nand-ecc-engine", 0);
+	if (!np)
+		return host;
+
+	ecc_pdev = of_find_device_by_node(np);
+	if (!ecc_pdev) {
+		of_node_put(np);
+		return NULL;
+	}
+
+	platform_device_put(ecc_pdev);
+	of_node_put(np);
+
+	return &ecc_pdev->dev;
+}
+
 MODULE_LICENSE("GPL");
 MODULE_AUTHOR("Miquel Raynal <miquel.raynal@bootlin.com>");
 MODULE_DESCRIPTION("Generic ECC engine");
diff --git a/drivers/mtd/nand/spi/core.c b/drivers/mtd/nand/spi/core.c
index 2c8685f1f2fa..ff8336870bc0 100644
--- a/drivers/mtd/nand/spi/core.c
+++ b/drivers/mtd/nand/spi/core.c
@@ -381,7 +381,10 @@ static int spinand_read_from_cache_op(struct spinand_device *spinand,
 		}
 	}
 
-	rdesc = spinand->dirmaps[req->pos.plane].rdesc;
+	if (req->mode == MTD_OPS_RAW)
+		rdesc = spinand->dirmaps[req->pos.plane].rdesc;
+	else
+		rdesc = spinand->dirmaps[req->pos.plane].rdesc_ecc;
 
 	while (nbytes) {
 		ret = spi_mem_dirmap_read(rdesc, column, nbytes, buf);
@@ -452,7 +455,10 @@ static int spinand_write_to_cache_op(struct spinand_device *spinand,
 			       req->ooblen);
 	}
 
-	wdesc = spinand->dirmaps[req->pos.plane].wdesc;
+	if (req->mode == MTD_OPS_RAW)
+		wdesc = spinand->dirmaps[req->pos.plane].wdesc;
+	else
+		wdesc = spinand->dirmaps[req->pos.plane].wdesc_ecc;
 
 	while (nbytes) {
 		ret = spi_mem_dirmap_write(wdesc, column, nbytes, buf);
@@ -865,6 +871,31 @@ static int spinand_create_dirmap(struct spinand_device *spinand,
 
 	spinand->dirmaps[plane].rdesc = desc;
 
+	if (nand->ecc.engine->integration != NAND_ECC_ENGINE_INTEGRATION_PIPELINED) {
+		spinand->dirmaps[plane].wdesc_ecc = spinand->dirmaps[plane].wdesc;
+		spinand->dirmaps[plane].rdesc_ecc = spinand->dirmaps[plane].rdesc;
+
+		return 0;
+	}
+
+	info.op_tmpl = *spinand->op_templates.update_cache;
+	info.op_tmpl.data.ecc = true;
+	desc = devm_spi_mem_dirmap_create(&spinand->spimem->spi->dev,
+					  spinand->spimem, &info);
+	if (IS_ERR(desc))
+		return PTR_ERR(desc);
+
+	spinand->dirmaps[plane].wdesc_ecc = desc;
+
+	info.op_tmpl = *spinand->op_templates.read_cache;
+	info.op_tmpl.data.ecc = true;
+	desc = devm_spi_mem_dirmap_create(&spinand->spimem->spi->dev,
+					  spinand->spimem, &info);
+	if (IS_ERR(desc))
+		return PTR_ERR(desc);
+
+	spinand->dirmaps[plane].rdesc_ecc = desc;
+
 	return 0;
 }
 
@@ -1208,14 +1239,6 @@ static int spinand_init(struct spinand_device *spinand)
 	if (ret)
 		goto err_free_bufs;
 
-	ret = spinand_create_dirmaps(spinand);
-	if (ret) {
-		dev_err(dev,
-			"Failed to create direct mappings for read/write operations (err = %d)\n",
-			ret);
-		goto err_manuf_cleanup;
-	}
-
 	ret = nanddev_init(nand, &spinand_ops, THIS_MODULE);
 	if (ret)
 		goto err_manuf_cleanup;
@@ -1250,6 +1273,14 @@ static int spinand_init(struct spinand_device *spinand)
 	mtd->ecc_strength = nanddev_get_ecc_conf(nand)->strength;
 	mtd->ecc_step_size = nanddev_get_ecc_conf(nand)->step_size;
 
+	ret = spinand_create_dirmaps(spinand);
+	if (ret) {
+		dev_err(dev,
+			"Failed to create direct mappings for read/write operations (err = %d)\n",
+			ret);
+		goto err_cleanup_ecc_engine;
+	}
+
 	return 0;
 
 err_cleanup_ecc_engine:
diff --git a/drivers/mtd/nand/spi/macronix.c b/drivers/mtd/nand/spi/macronix.c
index 3f31f1381a62..dce835132a1e 100644
--- a/drivers/mtd/nand/spi/macronix.c
+++ b/drivers/mtd/nand/spi/macronix.c
@@ -20,7 +20,7 @@ static SPINAND_OP_VARIANTS(read_cache_variants,
 
 static SPINAND_OP_VARIANTS(write_cache_variants,
 		SPINAND_PROG_LOAD_X4(true, 0, NULL, 0),
-		SPINAND_PROG_LOAD(true, 0, NULL, 0));
+		SPINAND_PROG_LOAD(false, 0, NULL, 0));
 
 static SPINAND_OP_VARIANTS(update_cache_variants,
 		SPINAND_PROG_LOAD_X4(false, 0, NULL, 0),
diff --git a/drivers/spi/Kconfig b/drivers/spi/Kconfig
index b2a8821971e1..269613b03e40 100644
--- a/drivers/spi/Kconfig
+++ b/drivers/spi/Kconfig
@@ -879,6 +879,7 @@ config SPI_SYNQUACER
 config SPI_MXIC
 	tristate "Macronix MX25F0A SPI controller"
 	depends on SPI_MASTER
+	imply MTD_NAND_ECC_MXIC
 	help
 	  This selects the Macronix MX25F0A SPI controller driver.
 
diff --git a/drivers/spi/spi-cadence-quadspi.c b/drivers/spi/spi-cadence-quadspi.c
index b808c94641fa..b0c9f62ccefb 100644
--- a/drivers/spi/spi-cadence-quadspi.c
+++ b/drivers/spi/spi-cadence-quadspi.c
@@ -1441,10 +1441,7 @@ static bool cqspi_supports_mem_op(struct spi_mem *mem,
 	if (!(all_true || all_false))
 		return false;
 
-	if (all_true)
-		return spi_mem_dtr_supports_op(mem, op);
-	else
-		return spi_mem_default_supports_op(mem, op);
+	return spi_mem_default_supports_op(mem, op);
 }
 
 static int cqspi_of_get_flash_pdata(struct platform_device *pdev,
@@ -1595,6 +1592,10 @@ static const struct spi_controller_mem_ops cqspi_mem_ops = {
 	.supports_op = cqspi_supports_mem_op,
 };
 
+static const struct spi_controller_mem_caps cqspi_mem_caps = {
+	.dtr = true,
+};
+
 static int cqspi_setup_flash(struct cqspi_st *cqspi)
 {
 	struct platform_device *pdev = cqspi->pdev;
@@ -1652,6 +1653,7 @@ static int cqspi_probe(struct platform_device *pdev)
 	}
 	master->mode_bits = SPI_RX_QUAD | SPI_RX_DUAL;
 	master->mem_ops = &cqspi_mem_ops;
+	master->mem_caps = &cqspi_mem_caps;
 	master->dev.of_node = pdev->dev.of_node;
 
 	cqspi = spi_master_get_devdata(master);
diff --git a/drivers/spi/spi-mem.c b/drivers/spi/spi-mem.c
index 37f4443ce9a0..f38ac31961c9 100644
--- a/drivers/spi/spi-mem.c
+++ b/drivers/spi/spi-mem.c
@@ -160,24 +160,28 @@ static bool spi_mem_check_buswidth(struct spi_mem *mem,
 	return true;
 }
 
-bool spi_mem_dtr_supports_op(struct spi_mem *mem,
-			     const struct spi_mem_op *op)
-{
-	if (op->cmd.nbytes != 2)
-		return false;
-
-	return spi_mem_check_buswidth(mem, op);
-}
-EXPORT_SYMBOL_GPL(spi_mem_dtr_supports_op);
-
 bool spi_mem_default_supports_op(struct spi_mem *mem,
 				 const struct spi_mem_op *op)
 {
-	if (op->cmd.dtr || op->addr.dtr || op->dummy.dtr || op->data.dtr)
-		return false;
+	struct spi_controller *ctlr = mem->spi->controller;
+	bool op_is_dtr =
+		op->cmd.dtr || op->addr.dtr || op->dummy.dtr || op->data.dtr;
 
-	if (op->cmd.nbytes != 1)
-		return false;
+	if (op_is_dtr) {
+		if (!spi_mem_controller_is_capable(ctlr, dtr))
+			return false;
+
+		if (op->cmd.nbytes != 2)
+			return false;
+	} else {
+		if (op->cmd.nbytes != 1)
+			return false;
+	}
+
+	if (op->data.ecc) {
+		if (!spi_mem_controller_is_capable(ctlr, ecc))
+			return false;
+	}
 
 	return spi_mem_check_buswidth(mem, op);
 }
diff --git a/drivers/spi/spi-mxic.c b/drivers/spi/spi-mxic.c
index 45889947afed..55c092069301 100644
--- a/drivers/spi/spi-mxic.c
+++ b/drivers/spi/spi-mxic.c
@@ -12,6 +12,8 @@
 #include <linux/io.h>
 #include <linux/iopoll.h>
 #include <linux/module.h>
+#include <linux/mtd/nand.h>
+#include <linux/mtd/nand-ecc-mxic.h>
 #include <linux/platform_device.h>
 #include <linux/pm_runtime.h>
 #include <linux/spi/spi.h>
@@ -167,11 +169,23 @@
 #define HW_TEST(x)		(0xe0 + ((x) * 4))
 
 struct mxic_spi {
+	struct device *dev;
 	struct clk *ps_clk;
 	struct clk *send_clk;
 	struct clk *send_dly_clk;
 	void __iomem *regs;
 	u32 cur_speed_hz;
+	struct {
+		void __iomem *map;
+		dma_addr_t dma;
+		size_t size;
+	} linear;
+
+	struct {
+		bool use_pipelined_conf;
+		struct nand_ecc_engine *pipelined_engine;
+		void *ctx;
+	} ecc;
 };
 
 static int mxic_spi_clk_enable(struct mxic_spi *mxic)
@@ -280,6 +294,51 @@ static void mxic_spi_hw_init(struct mxic_spi *mxic)
 	       mxic->regs + HC_CFG);
 }
 
+static u32 mxic_spi_prep_hc_cfg(struct spi_device *spi, u32 flags)
+{
+	int nio = 1;
+
+	if (spi->mode & (SPI_TX_OCTAL | SPI_RX_OCTAL))
+		nio = 8;
+	else if (spi->mode & (SPI_TX_QUAD | SPI_RX_QUAD))
+		nio = 4;
+	else if (spi->mode & (SPI_TX_DUAL | SPI_RX_DUAL))
+		nio = 2;
+
+	return flags | HC_CFG_NIO(nio) |
+	       HC_CFG_TYPE(spi->chip_select, HC_CFG_TYPE_SPI_NOR) |
+	       HC_CFG_SLV_ACT(spi->chip_select) | HC_CFG_IDLE_SIO_LVL(1);
+}
+
+static u32 mxic_spi_mem_prep_op_cfg(const struct spi_mem_op *op,
+				    unsigned int data_len)
+{
+	u32 cfg = OP_CMD_BYTES(op->cmd.nbytes) |
+		  OP_CMD_BUSW(fls(op->cmd.buswidth) - 1) |
+		  (op->cmd.dtr ? OP_CMD_DDR : 0);
+
+	if (op->addr.nbytes)
+		cfg |= OP_ADDR_BYTES(op->addr.nbytes) |
+		       OP_ADDR_BUSW(fls(op->addr.buswidth) - 1) |
+		       (op->addr.dtr ? OP_ADDR_DDR : 0);
+
+	if (op->dummy.nbytes)
+		cfg |= OP_DUMMY_CYC(op->dummy.nbytes);
+
+	/* Direct mapping data.nbytes field is not populated */
+	if (data_len) {
+		cfg |= OP_DATA_BUSW(fls(op->data.buswidth) - 1) |
+		       (op->data.dtr ? OP_DATA_DDR : 0);
+		if (op->data.dir == SPI_MEM_DATA_IN) {
+			cfg |= OP_READ;
+			if (op->data.dtr)
+				cfg |= OP_DQS_EN;
+		}
+	}
+
+	return cfg;
+}
+
 static int mxic_spi_data_xfer(struct mxic_spi *mxic, const void *txbuf,
 			      void *rxbuf, unsigned int len)
 {
@@ -304,25 +363,21 @@ static int mxic_spi_data_xfer(struct mxic_spi *mxic, const void *txbuf,
 
 		writel(data, mxic->regs + TXD(nbytes % 4));
 
+		ret = readl_poll_timeout(mxic->regs + INT_STS, sts,
+					 sts & INT_TX_EMPTY, 0, USEC_PER_SEC);
+		if (ret)
+			return ret;
+
+		ret = readl_poll_timeout(mxic->regs + INT_STS, sts,
+					 sts & INT_RX_NOT_EMPTY, 0,
+					 USEC_PER_SEC);
+		if (ret)
+			return ret;
+
+		data = readl(mxic->regs + RXD);
 		if (rxbuf) {
-			ret = readl_poll_timeout(mxic->regs + INT_STS, sts,
-						 sts & INT_TX_EMPTY, 0,
-						 USEC_PER_SEC);
-			if (ret)
-				return ret;
-
-			ret = readl_poll_timeout(mxic->regs + INT_STS, sts,
-						 sts & INT_RX_NOT_EMPTY, 0,
-						 USEC_PER_SEC);
-			if (ret)
-				return ret;
-
-			data = readl(mxic->regs + RXD);
 			data >>= (8 * (4 - nbytes));
 			memcpy(rxbuf + pos, &data, nbytes);
-			WARN_ON(readl(mxic->regs + INT_STS) & INT_RX_NOT_EMPTY);
-		} else {
-			readl(mxic->regs + RXD);
 		}
 		WARN_ON(readl(mxic->regs + INT_STS) & INT_RX_NOT_EMPTY);
 
@@ -332,11 +387,96 @@ static int mxic_spi_data_xfer(struct mxic_spi *mxic, const void *txbuf,
 	return 0;
 }
 
+static ssize_t mxic_spi_mem_dirmap_read(struct spi_mem_dirmap_desc *desc,
+					u64 offs, size_t len, void *buf)
+{
+	struct mxic_spi *mxic = spi_master_get_devdata(desc->mem->spi->master);
+	int ret;
+	u32 sts;
+
+	if (WARN_ON(offs + desc->info.offset + len > U32_MAX))
+		return -EINVAL;
+
+	writel(mxic_spi_prep_hc_cfg(desc->mem->spi, 0), mxic->regs + HC_CFG);
+
+	writel(mxic_spi_mem_prep_op_cfg(&desc->info.op_tmpl, len),
+	       mxic->regs + LRD_CFG);
+	writel(desc->info.offset + offs, mxic->regs + LRD_ADDR);
+	len = min_t(size_t, len, mxic->linear.size);
+	writel(len, mxic->regs + LRD_RANGE);
+	writel(LMODE_CMD0(desc->info.op_tmpl.cmd.opcode) |
+	       LMODE_SLV_ACT(desc->mem->spi->chip_select) |
+	       LMODE_EN,
+	       mxic->regs + LRD_CTRL);
+
+	if (mxic->ecc.use_pipelined_conf && desc->info.op_tmpl.data.ecc) {
+		ret = mxic_ecc_process_data_pipelined(mxic->ecc.pipelined_engine,
+						      NAND_PAGE_READ,
+						      mxic->linear.dma + offs);
+		if (ret)
+			return ret;
+	} else {
+		memcpy_fromio(buf, mxic->linear.map, len);
+	}
+
+	writel(INT_LRD_DIS, mxic->regs + INT_STS);
+	writel(0, mxic->regs + LRD_CTRL);
+
+	ret = readl_poll_timeout(mxic->regs + INT_STS, sts,
+				 sts & INT_LRD_DIS, 0, USEC_PER_SEC);
+	if (ret)
+		return ret;
+
+	return len;
+}
+
+static ssize_t mxic_spi_mem_dirmap_write(struct spi_mem_dirmap_desc *desc,
+					 u64 offs, size_t len,
+					 const void *buf)
+{
+	struct mxic_spi *mxic = spi_master_get_devdata(desc->mem->spi->master);
+	u32 sts;
+	int ret;
+
+	if (WARN_ON(offs + desc->info.offset + len > U32_MAX))
+		return -EINVAL;
+
+	writel(mxic_spi_prep_hc_cfg(desc->mem->spi, 0), mxic->regs + HC_CFG);
+
+	writel(mxic_spi_mem_prep_op_cfg(&desc->info.op_tmpl, len),
+	       mxic->regs + LWR_CFG);
+	writel(desc->info.offset + offs, mxic->regs + LWR_ADDR);
+	len = min_t(size_t, len, mxic->linear.size);
+	writel(len, mxic->regs + LWR_RANGE);
+	writel(LMODE_CMD0(desc->info.op_tmpl.cmd.opcode) |
+	       LMODE_SLV_ACT(desc->mem->spi->chip_select) |
+	       LMODE_EN,
+	       mxic->regs + LWR_CTRL);
+
+	if (mxic->ecc.use_pipelined_conf && desc->info.op_tmpl.data.ecc) {
+		ret = mxic_ecc_process_data_pipelined(mxic->ecc.pipelined_engine,
+						      NAND_PAGE_WRITE,
+						      mxic->linear.dma + offs);
+		if (ret)
+			return ret;
+	} else {
+		memcpy_toio(mxic->linear.map, buf, len);
+	}
+
+	writel(INT_LWR_DIS, mxic->regs + INT_STS);
+	writel(0, mxic->regs + LWR_CTRL);
+
+	ret = readl_poll_timeout(mxic->regs + INT_STS, sts,
+				 sts & INT_LWR_DIS, 0, USEC_PER_SEC);
+	if (ret)
+		return ret;
+
+	return len;
+}
+
 static bool mxic_spi_mem_supports_op(struct spi_mem *mem,
 				     const struct spi_mem_op *op)
 {
-	bool all_false;
-
 	if (op->data.buswidth > 8 || op->addr.buswidth > 8 ||
 	    op->dummy.buswidth > 8 || op->cmd.buswidth > 8)
 		return false;
@@ -348,64 +488,43 @@ static bool mxic_spi_mem_supports_op(struct spi_mem *mem,
 	if (op->addr.nbytes > 7)
 		return false;
 
-	all_false = !op->cmd.dtr && !op->addr.dtr && !op->dummy.dtr &&
-		    !op->data.dtr;
+	return spi_mem_default_supports_op(mem, op);
+}
+
+static int mxic_spi_mem_dirmap_create(struct spi_mem_dirmap_desc *desc)
+{
+	struct mxic_spi *mxic = spi_master_get_devdata(desc->mem->spi->master);
 
-	if (all_false)
-		return spi_mem_default_supports_op(mem, op);
-	else
-		return spi_mem_dtr_supports_op(mem, op);
+	if (!mxic->linear.map)
+		return -EINVAL;
+
+	if (desc->info.offset + desc->info.length > U32_MAX)
+		return -EINVAL;
+
+	if (!mxic_spi_mem_supports_op(desc->mem, &desc->info.op_tmpl))
+		return -EOPNOTSUPP;
+
+	return 0;
 }
 
 static int mxic_spi_mem_exec_op(struct spi_mem *mem,
 				const struct spi_mem_op *op)
 {
 	struct mxic_spi *mxic = spi_master_get_devdata(mem->spi->master);
-	int nio = 1, i, ret;
-	u32 ss_ctrl;
+	int i, ret;
 	u8 addr[8], cmd[2];
 
 	ret = mxic_spi_set_freq(mxic, mem->spi->max_speed_hz);
 	if (ret)
 		return ret;
 
-	if (mem->spi->mode & (SPI_TX_OCTAL | SPI_RX_OCTAL))
-		nio = 8;
-	else if (mem->spi->mode & (SPI_TX_QUAD | SPI_RX_QUAD))
-		nio = 4;
-	else if (mem->spi->mode & (SPI_TX_DUAL | SPI_RX_DUAL))
-		nio = 2;
-
-	writel(HC_CFG_NIO(nio) |
-	       HC_CFG_TYPE(mem->spi->chip_select, HC_CFG_TYPE_SPI_NOR) |
-	       HC_CFG_SLV_ACT(mem->spi->chip_select) | HC_CFG_IDLE_SIO_LVL(1) |
-	       HC_CFG_MAN_CS_EN,
+	writel(mxic_spi_prep_hc_cfg(mem->spi, HC_CFG_MAN_CS_EN),
 	       mxic->regs + HC_CFG);
-	writel(HC_EN_BIT, mxic->regs + HC_EN);
-
-	ss_ctrl = OP_CMD_BYTES(op->cmd.nbytes) |
-		  OP_CMD_BUSW(fls(op->cmd.buswidth) - 1) |
-		  (op->cmd.dtr ? OP_CMD_DDR : 0);
-
-	if (op->addr.nbytes)
-		ss_ctrl |= OP_ADDR_BYTES(op->addr.nbytes) |
-			   OP_ADDR_BUSW(fls(op->addr.buswidth) - 1) |
-			   (op->addr.dtr ? OP_ADDR_DDR : 0);
-
-	if (op->dummy.nbytes)
-		ss_ctrl |= OP_DUMMY_CYC(op->dummy.nbytes);
 
-	if (op->data.nbytes) {
-		ss_ctrl |= OP_DATA_BUSW(fls(op->data.buswidth) - 1) |
-			   (op->data.dtr ? OP_DATA_DDR : 0);
-		if (op->data.dir == SPI_MEM_DATA_IN) {
-			ss_ctrl |= OP_READ;
-			if (op->data.dtr)
-				ss_ctrl |= OP_DQS_EN;
-		}
-	}
+	writel(HC_EN_BIT, mxic->regs + HC_EN);
 
-	writel(ss_ctrl, mxic->regs + SS_CTRL(mem->spi->chip_select));
+	writel(mxic_spi_mem_prep_op_cfg(op, op->data.nbytes),
+	       mxic->regs + SS_CTRL(mem->spi->chip_select));
 
 	writel(readl(mxic->regs + HC_CFG) | HC_CFG_MAN_CS_ASSERT,
 	       mxic->regs + HC_CFG);
@@ -446,6 +565,14 @@ out:
 static const struct spi_controller_mem_ops mxic_spi_mem_ops = {
 	.supports_op = mxic_spi_mem_supports_op,
 	.exec_op = mxic_spi_mem_exec_op,
+	.dirmap_create = mxic_spi_mem_dirmap_create,
+	.dirmap_read = mxic_spi_mem_dirmap_read,
+	.dirmap_write = mxic_spi_mem_dirmap_write,
+};
+
+static const struct spi_controller_mem_caps mxic_spi_mem_caps = {
+	.dtr = true,
+	.ecc = true,
 };
 
 static void mxic_spi_set_cs(struct spi_device *spi, bool lvl)
@@ -510,6 +637,80 @@ static int mxic_spi_transfer_one(struct spi_master *master,
 	return 0;
 }
 
+/* ECC wrapper */
+static int mxic_spi_mem_ecc_init_ctx(struct nand_device *nand)
+{
+	struct nand_ecc_engine_ops *ops = mxic_ecc_get_pipelined_ops();
+	struct mxic_spi *mxic = nand->ecc.engine->priv;
+
+	mxic->ecc.use_pipelined_conf = true;
+
+	return ops->init_ctx(nand);
+}
+
+static void mxic_spi_mem_ecc_cleanup_ctx(struct nand_device *nand)
+{
+	struct nand_ecc_engine_ops *ops = mxic_ecc_get_pipelined_ops();
+	struct mxic_spi *mxic = nand->ecc.engine->priv;
+
+	mxic->ecc.use_pipelined_conf = false;
+
+	ops->cleanup_ctx(nand);
+}
+
+static int mxic_spi_mem_ecc_prepare_io_req(struct nand_device *nand,
+					   struct nand_page_io_req *req)
+{
+	struct nand_ecc_engine_ops *ops = mxic_ecc_get_pipelined_ops();
+
+	return ops->prepare_io_req(nand, req);
+}
+
+static int mxic_spi_mem_ecc_finish_io_req(struct nand_device *nand,
+					  struct nand_page_io_req *req)
+{
+	struct nand_ecc_engine_ops *ops = mxic_ecc_get_pipelined_ops();
+
+	return ops->finish_io_req(nand, req);
+}
+
+static struct nand_ecc_engine_ops mxic_spi_mem_ecc_engine_pipelined_ops = {
+	.init_ctx = mxic_spi_mem_ecc_init_ctx,
+	.cleanup_ctx = mxic_spi_mem_ecc_cleanup_ctx,
+	.prepare_io_req = mxic_spi_mem_ecc_prepare_io_req,
+	.finish_io_req = mxic_spi_mem_ecc_finish_io_req,
+};
+
+static void mxic_spi_mem_ecc_remove(struct mxic_spi *mxic)
+{
+	if (mxic->ecc.pipelined_engine) {
+		mxic_ecc_put_pipelined_engine(mxic->ecc.pipelined_engine);
+		nand_ecc_unregister_on_host_hw_engine(mxic->ecc.pipelined_engine);
+	}
+}
+
+static int mxic_spi_mem_ecc_probe(struct platform_device *pdev,
+				  struct mxic_spi *mxic)
+{
+	struct nand_ecc_engine *eng;
+
+	if (!mxic_ecc_get_pipelined_ops())
+		return -EOPNOTSUPP;
+
+	eng = mxic_ecc_get_pipelined_engine(pdev);
+	if (IS_ERR(eng))
+		return PTR_ERR(eng);
+
+	eng->dev = &pdev->dev;
+	eng->integration = NAND_ECC_ENGINE_INTEGRATION_PIPELINED;
+	eng->ops = &mxic_spi_mem_ecc_engine_pipelined_ops;
+	eng->priv = mxic;
+	mxic->ecc.pipelined_engine = eng;
+	nand_ecc_register_on_host_hw_engine(eng);
+
+	return 0;
+}
+
 static int __maybe_unused mxic_spi_runtime_suspend(struct device *dev)
 {
 	struct spi_master *master = dev_get_drvdata(dev);
@@ -555,6 +756,7 @@ static int mxic_spi_probe(struct platform_device *pdev)
 	platform_set_drvdata(pdev, master);
 
 	mxic = spi_master_get_devdata(master);
+	mxic->dev = &pdev->dev;
 
 	master->dev.of_node = pdev->dev.of_node;
 
@@ -575,11 +777,21 @@ static int mxic_spi_probe(struct platform_device *pdev)
 	if (IS_ERR(mxic->regs))
 		return PTR_ERR(mxic->regs);
 
+	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "dirmap");
+	mxic->linear.map = devm_ioremap_resource(&pdev->dev, res);
+	if (!IS_ERR(mxic->linear.map)) {
+		mxic->linear.dma = res->start;
+		mxic->linear.size = resource_size(res);
+	} else {
+		mxic->linear.map = NULL;
+	}
+
 	pm_runtime_enable(&pdev->dev);
 	master->auto_runtime_pm = true;
 
 	master->num_chipselect = 1;
 	master->mem_ops = &mxic_spi_mem_ops;
+	master->mem_caps = &mxic_spi_mem_caps;
 
 	master->set_cs = mxic_spi_set_cs;
 	master->transfer_one = mxic_spi_transfer_one;
@@ -591,6 +803,12 @@ static int mxic_spi_probe(struct platform_device *pdev)
 
 	mxic_spi_hw_init(mxic);
 
+	ret = mxic_spi_mem_ecc_probe(pdev, mxic);
+	if (ret == -EPROBE_DEFER) {
+		pm_runtime_disable(&pdev->dev);
+		return ret;
+	}
+
 	ret = spi_register_master(master);
 	if (ret) {
 		dev_err(&pdev->dev, "spi_register_master failed\n");
@@ -603,8 +821,10 @@ static int mxic_spi_probe(struct platform_device *pdev)
 static int mxic_spi_remove(struct platform_device *pdev)
 {
 	struct spi_master *master = platform_get_drvdata(pdev);
+	struct mxic_spi *mxic = spi_master_get_devdata(master);
 
 	pm_runtime_disable(&pdev->dev);
+	mxic_spi_mem_ecc_remove(mxic);
 	spi_unregister_master(master);
 
 	return 0;