[PATCH 2/4] mtd: devices: elm: Add support for ELM error correction

Peter Korsgaard jacmet at sunsite.dk
Mon Oct 15 15:40:47 EDT 2012


>>>>> Philip, Avinash <avinashphilip at ti.com> writes:

 > Platforms containing the ELM module can be used to correct errors
 > reported by BCH 4, 8 & 16 bit ECC scheme. For now only 4 & 8 bit
 > support is added.

This sounds odd to me. What about something like:

The ELM hardware module can be used to speedup BCH 4/8/16 ECC scheme
error correction.

For now only 4 & 8 bit support is added.


 > +++ b/drivers/mtd/devices/Makefile
 > @@ -17,8 +17,10 @@ obj-$(CONFIG_MTD_LART)		+= lart.o
 >  obj-$(CONFIG_MTD_BLOCK2MTD)	+= block2mtd.o
 >  obj-$(CONFIG_MTD_DATAFLASH)	+= mtd_dataflash.o
 >  obj-$(CONFIG_MTD_M25P80)	+= m25p80.o
 > +obj-$(CONFIG_MTD_NAND_OMAP2)	+= elm.o

You seem to only use it in 4/4 if CONFIG_MTD_NAND_OMAP_BCH is set, so it
probably makes more sense to use that symbol to not needlessly include
it if it won't be used.


 > +++ b/drivers/mtd/devices/elm.c
 > @@ -0,0 +1,440 @@
 > +/*
 > + * Error Location Module
 > + *
 > + * Copyright (C) 2012 Texas Instruments Incorporated - http://www.ti.com/
 > + *
 > + * This program is free software; you can redistribute it and/or modify
 > + * it under the terms of the GNU General Public License as published by
 > + * the Free Software Foundation; either version 2 of the License, or
 > + * (at your option) any later version.
 > + *
 > + * This program is distributed in the hope that it will be useful,
 > + * but WITHOUT ANY WARRANTY; without even the implied warranty of
 > + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
 > + * GNU General Public License for more details.
 > + *
 > + */
 > +
 > +#include <linux/platform_device.h>
 > +#include <linux/module.h>
 > +#include <linux/interrupt.h>
 > +#include <linux/io.h>
 > +#include <linux/of.h>
 > +#include <linux/pm_runtime.h>
 > +#include <linux/platform_data/elm.h>
 > +
 > +#define ELM_IRQSTATUS			0x018
 > +#define ELM_IRQENABLE			0x01c
 > +#define ELM_LOCATION_CONFIG		0x020
 > +#define ELM_PAGE_CTRL			0x080
 > +#define ELM_SYNDROME_FRAGMENT_0		0x400
 > +#define ELM_SYNDROME_FRAGMENT_6		0x418
 > +#define ELM_LOCATION_STATUS		0x800
 > +#define ELM_ERROR_LOCATION_0		0x880
 > +
 > +/* ELM Interrupt Status Register */
 > +#define INTR_STATUS_PAGE_VALID		BIT(8)
 > +
 > +/* ELM Interrupt Enable Register */
 > +#define INTR_EN_PAGE_MASK		BIT(8)
 > +
 > +/* ELM Location Configuration Register */
 > +#define ECC_BCH_LEVEL_MASK		0x3
 > +
 > +/* ELM syndrome */
 > +#define ELM_SYNDROME_VALID		BIT(16)
 > +
 > +/* ELM_LOCATION_STATUS Register */
 > +#define ECC_CORRECTABLE_MASK		BIT(8)
 > +#define ECC_NB_ERRORS_MASK		0x1f
 > +
 > +/* ELM_ERROR_LOCATION_0-15 Registers */
 > +#define ECC_ERROR_LOCATION_MASK		0x1fff
 > +
 > +#define ELM_ECC_SIZE			0x7ff
 > +
 > +#define SYNDROME_FRAGMENT_REG_SIZE	0x40
 > +#define ERROR_LOCATION_SIZE		0x100
 > +#define MAX_BCH_ELM_ERROR		16
 > +#define ELM_FRAGMENT_REG		7
 > +
 > +typedef u32 syn_t[ELM_FRAGMENT_REG];
 > +typedef u32 elm_error_t[MAX_BCH_ELM_ERROR];
 > +
 > +struct elm_info {
 > +	struct device *dev;
 > +	void __iomem *elm_base;
 > +	struct completion elm_completion;
 > +	struct list_head list;
 > +	enum bch_ecc bch_type;
 > +};
 > +
 > +static LIST_HEAD(elm_devices);
 > +
 > +static void elm_write_reg(void *offset, u32 val)
 > +{
 > +	writel(val, offset);
 > +}
 > +
 > +static u32 elm_read_reg(void *offset)
 > +{
 > +	return readl(offset);
 > +}

As written these read/write wrappers don't add anything. How about
passing struct elm_info and offset as an integer so you can drop
elm_base from all call sites, E.G.:

static void elm_write_reg(struct elm_info *info, int offset, u32 val)
{
        writel(val, info->elm_base + offset);
}


 > +
 > +/**
 > + * elm_config - Configure ELM module
 > + * @info:	elm info
 > + */
 > +static void elm_config(struct elm_info *info)
 > +{
 > +	u32 reg_val;
 > +
 > +	reg_val = (info->bch_type & ECC_BCH_LEVEL_MASK) | (ELM_ECC_SIZE << 16);
 > +	elm_write_reg(info->elm_base + ELM_LOCATION_CONFIG, reg_val);
 > +}
 > +
 > +/**
 > + * elm_configure_page_mode - Enable/Disable page mode
 > + * @info:	elm info
 > + * @index:	index number of syndrome fragment vector
 > + * @enable:	enable/disable flag for page mode
 > + *
 > + * Enable page mode for syndrome fragment index
 > + */
 > +static void elm_configure_page_mode(struct elm_info *info, int index,
 > +		bool enable)
 > +{
 > +	u32 reg_val;
 > +
 > +	reg_val = elm_read_reg(info->elm_base + ELM_PAGE_CTRL);
 > +	if (enable)
 > +		reg_val |= BIT(index);	/* enable page mode */
 > +	else
 > +		reg_val &= ~BIT(index);	/* disable page mode */
 > +
 > +	elm_write_reg(info->elm_base + ELM_PAGE_CTRL, reg_val);
 > +}
 > +
 > +static void rearrange_bch4(u8 *syndrome)
 > +{
 > +	/*
 > +	 * BCH4 has 52 bit used for ecc, but OOB stored with
 > +	 * nibble 0 appended, removes appended 0 nibble
 > +	 */
 > +	u64 *dst = (u64 *)syndrome;
 > +	*dst >>= 4;
 > +}
 > +
 > +/**
 > + * elm_reverse_eccdata - Reverse ecc data
 > + * @info:	elm info
 > + * @ecc_data:	buffer for calculated ecc
 > + * @syndrome:	buffer for syndrome polynomial
 > + *
 > + * ecc_data has to be reversed for syndrome vector
 > + */
 > +static void elm_reverse_eccdata(struct elm_info *info, u8 *ecc_data,
 > +		u8 *syndrome)
 > +{
 > +	int i;
 > +	int bch_size = info->bch_type ? BCH8_ECC_OOB_BYTES : BCH4_SIZE;
 > +
 > +	for (i = 0; i < bch_size; i++)
 > +		syndrome[bch_size - 1 - i] = ecc_data[i];
 > +
 > +	/*
 > +	 * syndrome polynomial to be rearranged for BCH 4 ecc scheme as 52
 > +	 * bits being used and 4 bits being appended in syndrome vector
 > +	 */
 > +	if (info->bch_type == BCH4_ECC)
 > +		rearrange_bch4(syndrome);
 > +}
 > +
 > +/**
 > + * elm_load_syndrome - Load ELM syndrome reg
 > + * @info:	elm info
 > + * @index:	syndrome fragment index
 > + * @syndrome:	Syndrome polynomial
 > + *
 > + * Load syndrome fragment registers with syndrome polynomial
 > + */
 > +static void elm_load_syndrome(struct elm_info *info, int index, u8 *syndrome)
 > +{
 > +	int i;
 > +	int max = info->bch_type ? BCH8_SYNDROME_SIZE : BCH4_SYNDROME_SIZE;
 > +	void *syn_frag_base = info->elm_base + ELM_SYNDROME_FRAGMENT_0;
 > +	syn_t *syn_fragment;
 > +	u32 reg_val;
 > +
 > +	elm_configure_page_mode(info, index, true);
 > +	syn_fragment = syn_frag_base + SYNDROME_FRAGMENT_REG_SIZE * index;
 > +
 > +	for (i = 0; i < max; i++) {
 > +		reg_val = syndrome[0] | syndrome[1] << 8 | syndrome[2] << 16 |
 > +			syndrome[3] << 24;
 > +		elm_write_reg(*syn_fragment + i, reg_val);
 > +		/* Update syndrome polynomial pointer */
 > +		syndrome += 4;
 > +	}
 > +}
 > +
 > +/**
 > + * elm_start_processing - start elm syndrome processing
 > + * @info:	elm info
 > + * @err_vec:	elm error vectors
 > + *
 > + * Set syndrome valid bit for syndrome fragment registers for which
 > + * elm syndrome fragment registers are loaded. This enables elm module
 > + * to start processing syndrome vectors.
 > + */
 > +static void elm_start_processing(struct elm_info *info,
 > +		struct elm_errorvec *err_vec)
 > +{
 > +	int i;
 > +	void *offset;
 > +	u32 reg_val;
 > +
 > +	/*
 > +	 * Set syndrome vector valid so that ELM module will process it for
 > +	 * vectors error is reported
 > +	 */
 > +	for (i = 0; i < ERROR_VECTOR_MAX; i++) {
 > +		if (err_vec[i].error_reported) {
 > +			offset = info->elm_base + ELM_SYNDROME_FRAGMENT_6 + i *
 > +				SYNDROME_FRAGMENT_REG_SIZE;
 > +			reg_val = elm_read_reg(offset);
 > +			reg_val |= ELM_SYNDROME_VALID;
 > +			elm_write_reg(offset, reg_val);
 > +		}
 > +	}
 > +}
 > +
 > +/**
 > + * elm_error_correction - locate correctable error position
 > + * @info:	elm info
 > + * @err_vec:	elm error vectors
 > + *
 > + * On completion of processing by elm module, error location status
 > + * register updated with correctable/uncorrectable error information.
 > + * In case of correctable errors, number of errors located from
 > + * elm location status register & read the these many positions from
 > + * elm error location register.
 > + */
 > +static void elm_error_correction(struct elm_info *info,
 > +		struct elm_errorvec *err_vec)
 > +{
 > +	int i, j, errors = 0;
 > +	void *err_loc_base = info->elm_base + ELM_ERROR_LOCATION_0;
 > +	elm_error_t *err_loc;
 > +	void *offset;
 > +	u32 reg_val;
 > +
 > +	for (i = 0; i < ERROR_VECTOR_MAX; i++) {
 > +		/* check error reported */
 > +		if (err_vec[i].error_reported) {
 > +			offset = info->elm_base + ELM_LOCATION_STATUS +
 > +				i * ERROR_LOCATION_SIZE;
 > +			reg_val = elm_read_reg(offset);
 > +			/* check correctable error or not */
 > +			if (reg_val & ECC_CORRECTABLE_MASK) {
 > +				err_loc = err_loc_base +
 > +					i * ERROR_LOCATION_SIZE;
 > +				/* read count of correctable errors */
 > +				err_vec[i].error_count = reg_val &
 > +					ECC_NB_ERRORS_MASK;
 > +
 > +				/* update the error locations in error vector */
 > +				for (j = 0; j < err_vec[i].error_count; j++) {
 > +
 > +					reg_val = elm_read_reg(*err_loc + j);
 > +					err_vec[i].error_loc[j] = reg_val &
 > +						ECC_ERROR_LOCATION_MASK;
 > +				}
 > +
 > +				errors += err_vec[i].error_count;
 > +			} else {
 > +				err_vec[i].error_uncorrectable++;
 > +			}
 > +
 > +			/* clearing interrupts for processed error vectors */
 > +			elm_write_reg(info->elm_base + ELM_IRQSTATUS, BIT(i));
 > +
 > +			/* disable page mode */
 > +			elm_configure_page_mode(info, i, false);
 > +		}
 > +	}
 > +
 > +	return;
 > +}
 > +
 > +/**
 > + * elm_decode_bch_error_page - Locate error position
 > + * @info:	elm info
 > + * @ecc_calc:	calculated ECC bytes from GPMC
 > + * @err_vec:	elm error vectors
 > + *
 > + * Called with one or greater reported and is vectors with error reported
 > + * is updated in err_vec[].error_reported
 > + */
 > +void elm_decode_bch_error_page(struct device *dev, u8 *ecc_calc,
 > +		struct elm_errorvec *err_vec)
 > +{
 > +	int i;
 > +	u8 syndrome[BCH_MAX_ECC_BYTES_PER_SECTOR] = {0}, *syn_p;


Why do you need to keep the entire syndrome around? You seem to only use
it between elm_reverse_eccdata() and elm_load_syndrome(), so it could as
well be BCH8_ECC_OOB_BYTES long (or rather a multiple of sizeof(u32).

It would also be good to do the shuffeling directly in elm_load_syndrome
so you don't need the extra copy.


 > +	struct elm_info *info = dev_get_drvdata(dev);
 > +	u32 reg_val;
 > +
 > +	/* enable page mode interrupt */
 > +	reg_val = elm_read_reg(info->elm_base + ELM_IRQSTATUS);
 > +	elm_write_reg(info->elm_base + ELM_IRQSTATUS,
 > +			reg_val & INTR_STATUS_PAGE_VALID);
 > +	elm_write_reg(info->elm_base + ELM_IRQENABLE, INTR_EN_PAGE_MASK);
 > +
 > +	syn_p = syndrome;
 > +	for (i = 0; i < ERROR_VECTOR_MAX; i++) {
 > +		if (err_vec[i].error_reported) {
 > +			elm_reverse_eccdata(info, ecc_calc, syn_p);
 > +			elm_load_syndrome(info, i, syn_p);
 > +		}
 > +
 > +		ecc_calc += info->bch_type ? BCH8_SIZE : BCH4_SIZE;
 > +		syn_p += OOB_SECTOR_SIZE;
 > +	}
 > +
 > +	elm_start_processing(info, err_vec);
 > +
 > +	/*
 > +	 * In case of error reported, wait for ELM module to finish
 > +	 * locating error correction
 > +	 */
 > +	wait_for_completion(&info->elm_completion);
 > +
 > +	/* disable page mode interrupt */
 > +	reg_val = elm_read_reg(info->elm_base + ELM_IRQENABLE);
 > +	elm_write_reg(info->elm_base + ELM_IRQENABLE,
 > +			reg_val & ~INTR_EN_PAGE_MASK);
 > +	elm_error_correction(info, err_vec);
 > +}
 > +EXPORT_SYMBOL(elm_decode_bch_error_page);
 > +
 > +static irqreturn_t elm_isr(int this_irq, void *dev_id)
 > +{
 > +	u32 reg_val;
 > +	struct elm_info *info = dev_id;
 > +
 > +	reg_val = elm_read_reg(info->elm_base + ELM_IRQSTATUS);
 > +
 > +	/* all error vectors processed */
 > +	if (reg_val & INTR_STATUS_PAGE_VALID) {
 > +		elm_write_reg(info->elm_base + ELM_IRQSTATUS,
 > +				reg_val & INTR_STATUS_PAGE_VALID);
 > +		complete(&info->elm_completion);
 > +		return IRQ_HANDLED;
 > +	}
 > +
 > +	return IRQ_NONE;
 > +}
 > +
 > +struct device *elm_request(enum bch_ecc bch_type)
 > +{
 > +	struct elm_info *info;
 > +
 > +	list_for_each_entry(info, &elm_devices, list) {
 > +		if (info && info->dev) {
 > +				info->bch_type = bch_type;
 > +				elm_config(info);
 > +				return info->dev;
 > +		}
 > +	}
 > +
 > +	return NULL;
 > +}
 > +EXPORT_SYMBOL(elm_request);
 > +
 > +static int __devinit elm_probe(struct platform_device *pdev)
 > +{
 > +	int ret = 0;
 > +	struct resource *res, *irq;
 > +	struct elm_info *info;
 > +
 > +	info = devm_kzalloc(&pdev->dev, sizeof(*info), GFP_KERNEL);
 > +	if (!info) {
 > +		dev_err(&pdev->dev, "failed to allocate memory\n");
 > +		return -ENOMEM;
 > +	}
 > +
 > +	info->dev = &pdev->dev;
 > +
 > +	irq = platform_get_resource(pdev, IORESOURCE_IRQ, 0);
 > +	if (!irq) {
 > +		dev_err(&pdev->dev, "no irq resource defined\n");
 > +		return -ENODEV;
 > +	}
 > +
 > +	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
 > +	if (!res) {
 > +		dev_err(&pdev->dev, "no memory resource defined\n");
 > +		return -ENODEV;
 > +	}
 > +
 > +
 > +	info->elm_base = devm_request_and_ioremap(&pdev->dev, res);
 > +	if (!info->elm_base)
 > +		return -EADDRNOTAVAIL;
 > +
 > +
 > +	ret = devm_request_irq(&pdev->dev, irq->start, elm_isr, 0,
 > +			pdev->name, info);
 > +	if (ret) {
 > +		dev_err(&pdev->dev, "failure requesting irq %i\n", irq->start);
 > +		return ret;
 > +	}
 > +
 > +	pm_runtime_enable(&pdev->dev);
 > +	if (pm_runtime_get_sync(&pdev->dev)) {
 > +		ret = -EINVAL;
 > +		pm_runtime_disable(&pdev->dev);
 > +		dev_err(&pdev->dev, "can't enable clock\n");
 > +		return ret;
 > +	}
 > +
 > +	init_completion(&info->elm_completion);
 > +	INIT_LIST_HEAD(&info->list);
 > +	list_add(&info->list, &elm_devices);
 > +	platform_set_drvdata(pdev, info);
 > +	return ret;
 > +}
 > +
 > +static int __devexit elm_remove(struct platform_device *pdev)
 > +{
 > +	pm_runtime_put_sync(&pdev->dev);
 > +	pm_runtime_disable(&pdev->dev);
 > +	platform_set_drvdata(pdev, NULL);
 > +	return 0;
 > +}
 > +
 > +
 > +#ifdef CONFIG_OF
 > +static const struct of_device_id elm_of_match[] = {
 > +	{ .compatible = "ti,elm" },
 > +	{},
 > +};
 > +MODULE_DEVICE_TABLE(of, elm_of_match);
 > +#endif
 > +
 > +static struct platform_driver elm_driver = {
 > +	.driver		= {
 > +		.name	= "elm",
 > +		.of_match_table = of_match_ptr(elm_of_match),
 > +		.owner	= THIS_MODULE,
 > +	},
 > +	.probe		= elm_probe,
 > +	.remove		= __devexit_p(elm_remove),
 > +};
 > +
 > +module_platform_driver(elm_driver);
 > +
 > +MODULE_DESCRIPTION("ELM driver for BCH error correction");
 > +MODULE_AUTHOR("Texas Instruments");
 > +MODULE_ALIAS("platform: elm");
 > +MODULE_LICENSE("GPL v2");
 > diff --git a/include/linux/platform_data/elm.h b/include/linux/platform_data/elm.h
 > new file mode 100644
 > index 0000000..eb53163
 > --- /dev/null
 > +++ b/include/linux/platform_data/elm.h
 > @@ -0,0 +1,60 @@
 > +/*
 > + * BCH Error Location Module
 > + *
 > + * Copyright (C) 2012 Texas Instruments Incorporated - http://www.ti.com/
 > + *
 > + * This program is free software; you can redistribute it and/or modify
 > + * it under the terms of the GNU General Public License as published by
 > + * the Free Software Foundation; either version 2 of the License, or
 > + * (at your option) any later version.
 > + *
 > + * This program is distributed in the hope that it will be useful,
 > + * but WITHOUT ANY WARRANTY; without even the implied warranty of
 > + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
 > + * GNU General Public License for more details.
 > + *
 > + */
 > +
 > +#ifndef __ELM_H
 > +#define __ELM_H
 > +
 > +enum bch_ecc {
 > +	BCH4_ECC = 0,
 > +	BCH8_ECC,
 > +	BCH16_ECC,

It probably makes more sense to not provide the enum value for BCH16 as
you don't support it.


 > +};
 > +
 > +/* ELM support 8 error syndrome process */
 > +#define ERROR_VECTOR_MAX		8
 > +#define OOB_SECTOR_SIZE			16
 > +
 > +#define BCH_MAX_ECC_BYTES_PER_SECTOR	(OOB_SECTOR_SIZE * ERROR_VECTOR_MAX)
 > +
 > +#define BCH8_ECC_OOB_BYTES		13
 > +/* RBL requires 14 byte even though BCH8 uses only 13 byte */
 > +#define BCH8_SIZE			(BCH8_ECC_OOB_BYTES + 1)
 > +#define BCH4_SIZE			7
 > +
 > +#define	BCH8_SYNDROME_SIZE		4	/* 13 bytes of ecc */
 > +#define	BCH4_SYNDROME_SIZE		2	/* 7 bytes of ecc */
 > +
 > +/**
 > + * struct elm_errorvec - error vector for elm
 > + * @error_reported:		set true for vectors error is reported
 > + *
 > + * @error_count:		number of correctable errors in the sector
 > + * @error_uncorrectable:	number of uncorrectable errors
 > + * @error_loc:			buffer for error location
 > + *
 > + */
 > +struct elm_errorvec {
 > +	bool error_reported;
 > +	int error_count;
 > +	int error_uncorrectable;
 > +	int error_loc[ERROR_VECTOR_MAX];
 > +};
 > +
 > +void elm_decode_bch_error_page(struct device *dev, u8 *ecc_calc,
 > +		struct elm_errorvec *err_vec);
 > +struct device *elm_request(enum bch_ecc bch_type);
 > +#endif /* __ELM_H */
 > -- 
 > 1.7.0.4


 > ______________________________________________________
 > Linux MTD discussion mailing list
 > http://lists.infradead.org/mailman/listinfo/linux-mtd/


-- 
Bye, Peter Korsgaard



More information about the linux-mtd mailing list