lkml.org 
[lkml]   [2018]   [Apr]   [6]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [LINUX PATCH v8 2/2] mtd: rawnand: pl353: Add basic driver for arm pl353 smc nand interface
From
Date


On 03/23/2018 09:58 AM, Naga Sureshkumar Relli wrote:
> Hi Miquel,
>
> Thanks for reviewing the patch.
> Please see my comments inline.
>
>> -----Original Message-----
>> From: Miquel Raynal [mailto:miquel.raynal@bootlin.com]
>> Sent: Tuesday, March 20, 2018 4:08 AM
>> To: nagasureshkumarrelli@gmail.com
>> Cc: boris.brezillon@bootlin.com; richard@nod.at; dwmw2@infradead.org;
>> computersforpeace@gmail.com; marek.vasut@gmail.com;
>> cyrille.pitchen@wedev4u.fr; linux-mtd@lists.infradead.org; linux-
>> kernel@vger.kernel.org; Michal Simek <michals@xilinx.com>; Punnaiah
>> Choudary Kalluri <punnaia@xilinx.com>; Naga Sureshkumar Relli
>> <nagasure@xilinx.com>
>> Subject: Re: [LINUX PATCH v8 2/2] mtd: rawnand: pl353: Add basic driver for
>> arm pl353 smc nand interface
>>
>> Hi Naga,
>>
>> Thanks for sending a new version supporting ->exec_op(). A few comments
>> below.
>>
>> On Wed, 14 Mar 2018 16:18:25 +0530,
>> <nagasureshkumarrelli@gmail.com> wrote:
>>
>>> From: Naga Sureshkumar Relli <nagasure@xilinx.com>
>>>
>>> Add driver for arm pl353 static memory controller nand interface with
>>> HW ECC support. This controller is used in xilinx zynq soc for
>>> interfacing the nand flash memory.
>>>
>>> Signed-off-by: Naga Sureshkumar Relli <nagasure@xilinx.com>
>>> ---
>>> Changes in v8:
>>> - Added exec_op() implementation
>>> - Fixed the below v7 review comments
>>> - removed mtd_info from pl353_nand_info struct
>>> - Corrected ecc layout offsets
>>> - Added on-die ecc support
>>> Changes in v7:
>>> - Currently not implemented the memclk rate adjustments. I will
>>> look into this later and once the basic driver is accepted.
>>> - Fixed GPL licence ident
>>> Changes in v6:
>>> - Fixed the checkpatch.pl reported warnings
>>> - Using the address cycles information from the onfi param page
>>> earlier it is hardcoded to 5 in driver Changes in v5:
>>> - Configure the nand timing parameters as per the onfi spec Changes
>>> in v4:
>>> - Updated the driver to sync with pl353_smc driver APIs Changes in
>>> v3:
>>> - implemented the proper error codes
>>> - further breakdown this patch to multiple sets
>>> - added the controller and driver details to Documentation section
>>> - updated the licenece to GPLv2
>>> - reorganized the pl353_nand_ecc_init function Changes in v2:
>>> - use "depends on" rather than "select" option in kconfig
>>> - remove unused variable parts
>>> - remove dummy helper and use writel_relaxed directly
>>> ---
>>> drivers/mtd/nand/raw/Kconfig | 8 +
>>> drivers/mtd/nand/raw/Makefile | 1 +
>>> drivers/mtd/nand/raw/pl353_nand.c | 1363
>>> +++++++++++++++++++++++++++++++++++++
>>> 3 files changed, 1372 insertions(+)
>>> create mode 100644 drivers/mtd/nand/raw/pl353_nand.c
>>>
>>> diff --git a/drivers/mtd/nand/raw/Kconfig
>>> b/drivers/mtd/nand/raw/Kconfig index 2c6ecb7..5e20391 100644
>>> --- a/drivers/mtd/nand/raw/Kconfig
>>> +++ b/drivers/mtd/nand/raw/Kconfig
>>> @@ -566,4 +566,12 @@ config MTD_NAND_MTK
>>> Enables support for NAND controller on MTK SoCs.
>>> This controller is found on mt27xx, mt81xx, mt65xx SoCs.
>>>
>>> +config MTD_NAND_PL353
>>> + tristate "ARM Pl353 NAND flash driver"
>>> + depends on MTD_NAND && ARM
>>> + depends on PL35X_SMC
>>> + help
>>> + This enables access to the NAND flash device on PL353
>>> + SMC controller.
>>
>> What about:
>> "Enables support for PrimeCell Static Memory Controller
>> PL353."?
> Are you asking to change the description in help?
>>
>>> +
>>> endif # MTD_NAND
>>> diff --git a/drivers/mtd/nand/raw/Makefile
>>> b/drivers/mtd/nand/raw/Makefile index f16f59a..3e943f3 100644
>>> --- a/drivers/mtd/nand/raw/Makefile
>>> +++ b/drivers/mtd/nand/raw/Makefile
>>> @@ -57,6 +57,7 @@ obj-$(CONFIG_MTD_NAND_HISI504) +=
>> hisi504_nand.o
>>> obj-$(CONFIG_MTD_NAND_BRCMNAND) += brcmnand/
>>> obj-$(CONFIG_MTD_NAND_QCOM) += qcom_nandc.o
>>> obj-$(CONFIG_MTD_NAND_MTK) += mtk_ecc.o mtk_nand.o
>>> +obj-$(CONFIG_MTD_NAND_PL353) += pl353_nand.o
>>>
>>> nand-objs := nand_base.o nand_bbt.o nand_timings.o nand_ids.o
>>> nand-objs += nand_amd.o diff --git a/drivers/mtd/nand/raw/pl353_nand.c
>>> b/drivers/mtd/nand/raw/pl353_nand.c
>>> new file mode 100644
>>> index 0000000..55c51e2
>>> --- /dev/null
>>> +++ b/drivers/mtd/nand/raw/pl353_nand.c
>>> @@ -0,0 +1,1363 @@
>>> +// SPDX-License-Identifier: GPL-2.0
>>> +/*
>>> + * ARM PL353 NAND flash controller driver
>>> + *
>>> + * Copyright (C) 2017 Xilinx, Inc
>>> + * Author: Punnaiah <punnaiah@xilinx.com>
>>> + * Author: nagasuresh <nagasure@xilinx.com>
>>
>> Please use your full names here.
> Ok, I will update.
>>
>>> + *
>>> + */
>>> +
>>> +#include <linux/err.h>
>>> +#include <linux/delay.h>
>>> +#include <linux/interrupt.h>
>>> +#include <linux/io.h>
>>> +#include <linux/ioport.h>
>>> +#include <linux/irq.h>
>>> +#include <linux/module.h>
>>> +#include <linux/moduleparam.h>
>>> +#include <linux/mtd/mtd.h>
>>> +#include <linux/mtd/rawnand.h>
>>> +#include <linux/mtd/nand_ecc.h>
>>> +#include <linux/mtd/partitions.h>
>>> +#include <linux/of_address.h>
>>> +#include <linux/of_device.h>
>>> +#include <linux/of_platform.h>
>>> +#include <linux/platform_device.h>
>>> +#include <linux/slab.h>
>>> +#include <linux/platform_data/pl353-smc.h>
>>> +
>>> +#define PL353_NAND_DRIVER_NAME "pl353-nand"
>>> +
>>> +/* NAND flash driver defines */
>>> +#define PL353_NAND_CMD_PHASE 1 /* End command valid in
>> command phase */
>>> +#define PL353_NAND_DATA_PHASE 2 /* End command valid in data
>> phase */
>>> +#define PL353_NAND_ECC_SIZE 512 /* Size of data for ECC
>> operation */
>>> +
>>> +/* Flash memory controller operating parameters */
>>> +
>>> +#define PL353_NAND_ECC_CONFIG (BIT(4) | /* ECC read at end of
>> page */ \
>>> + (0 << 5)) /* No Jumping */
>>> +
>>> +/* AXI Address definitions */
>>> +#define START_CMD_SHIFT 3
>>> +#define END_CMD_SHIFT 11
>>> +#define END_CMD_VALID_SHIFT 20
>>> +#define ADDR_CYCLES_SHIFT 21
>>> +#define CLEAR_CS_SHIFT 21
>>> +#define ECC_LAST_SHIFT 10
>>> +#define COMMAND_PHASE (0 << 19)
>>> +#define DATA_PHASE BIT(19)
>>> +
>>> +#define PL353_NAND_ECC_LAST BIT(ECC_LAST_SHIFT) /* Set
>> ECC_Last */
>>> +#define PL353_NAND_CLEAR_CS BIT(CLEAR_CS_SHIFT) /* Clear chip
>> select */
>>> +
>>> +#define ONDIE_ECC_FEATURE_ADDR 0x90
>>> +#define PL353_NAND_ECC_BUSY_TIMEOUT (1 * HZ)
>>> +#define PL353_NAND_DEV_BUSY_TIMEOUT (1 * HZ)
>>> +#define PL353_NAND_LAST_TRANSFER_LENGTH 4
>>> +
>>> +/* Inline function for the NAND controller register write */ static
>>> +inline void pl353_nand_write32(void __iomem *addr, u32 val) {
>>> + writel_relaxed((val), (addr));
>>> +}
>>
>> Is there an actual need for this inline function? Why not calling
>> writel_relaxed() directly?
> Let me check and get back to you.
>>
>>> +
>>> +struct pl353_nfc_op {
>>> + u32 cmnds[4];
>>> + u32 thirdrow;
>>> + u32 type;
>>> + u32 end_cmd;
>>> + u32 addrs;
>>> + bool wait;
>>> + u32 len;
>>> + u32 naddrs;
>>> + unsigned int data_instr_idx;
>>> + const struct nand_op_instr *data_instr;
>>> + unsigned int rdy_timeout_ms;
>>> + unsigned int rdy_delay_ns;
>>> +};
>>> +
>>> +/**
>>> + * struct pl353_nand_info - Defines the NAND flash driver instance
>>> + * @chip: NAND chip information structure
>>> + * @nand_base: Virtual address of the NAND flash device
>>> + * @end_cmd_pending: End command is pending
>>> + * @end_cmd: End command
>>> + * @row_addr_cycles: Row address cycles
>>> + * @col_addr_cycles: Column address cycles
>>> + * @address: Page address
>>> + * @cmd_pending: More command is needed
>>> + */
>>> +struct pl353_nand_info {
>>> + struct nand_chip chip;
>>> + void __iomem *nand_base;
>>> + unsigned long end_cmd_pending;
>>> + unsigned long end_cmd;
>>> + u8 row_addr_cycles;
>>> + u8 col_addr_cycles;
>>> + u32 address;
>>> + u32 cmd_pending;
>>> +};
>>> +
>>> +static int pl353_ecc_ooblayout16_ecc(struct mtd_info *mtd, int section,
>>> + struct mtd_oob_region *oobregion) {
>>> + struct nand_chip *chip = mtd_to_nand(mtd);
>>> +
>>> + if (section >= chip->ecc.steps)
>>> + return -ERANGE;
>>> +
>>> + oobregion->offset = (section * 16) + 0;
>>> + oobregion->length = chip->ecc.bytes;
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +static int pl353_ecc_ooblayout16_free(struct mtd_info *mtd, int section,
>>> + struct mtd_oob_region *oobregion) {
>>> + struct nand_chip *chip = mtd_to_nand(mtd);
>>> +
>>> + if (section >= chip->ecc.steps)
>>> + return -ERANGE;
>>> +
>>> + oobregion->offset = (section * 16) + 8;
>>> +
>>
>> Extra space
> I will fix it.
>>
>>> + oobregion->length = 8;
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +static const struct mtd_ooblayout_ops pl353_ecc_ooblayout16_ops = {
>>> + .ecc = pl353_ecc_ooblayout16_ecc,
>>> + .free = pl353_ecc_ooblayout16_free,
>>> +};
>>> +
>>> +static int pl353_ecc_ooblayout64_ecc(struct mtd_info *mtd, int section,
>>> + struct mtd_oob_region *oobregion) {
>>> + struct nand_chip *chip = mtd_to_nand(mtd);
>>> +
>>> + if (section >= chip->ecc.steps)
>>> + return -ERANGE;
>>> +
>>> + oobregion->offset = (section * chip->ecc.bytes) + 52;
>>> + oobregion->length = chip->ecc.bytes;
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +static int pl353_ecc_ooblayout64_free(struct mtd_info *mtd, int section,
>>> + struct mtd_oob_region *oobregion) {
>>> + struct nand_chip *chip = mtd_to_nand(mtd);
>>> +
>>> + if (section)
>>> + return -ERANGE;
>>> + if (section >= chip->ecc.steps)
>>> + return -ERANGE;
>>
>> If you can't support more than one section, the second if is useless, and the
>> offset is just "2".
> Yes, we can use just 2. I will update
>>
>>> +
>>> + oobregion->offset = (section * chip->ecc.bytes) + 2;
>>> +
>>> + oobregion->length = 50;
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +static const struct mtd_ooblayout_ops pl353_ecc_ooblayout64_ops = {
>>> + .ecc = pl353_ecc_ooblayout64_ecc,
>>> + .free = pl353_ecc_ooblayout64_free,
>>> +};
>>> +
>>> +/* Generic flash bbt decriptors */
>>> +static uint8_t bbt_pattern[] = { 'B', 'b', 't', '0' }; static uint8_t
>>> +mirror_pattern[] = { '1', 't', 'b', 'B' };
>>> +
>>> +static struct nand_bbt_descr bbt_main_descr = {
>>> + .options = NAND_BBT_LASTBLOCK | NAND_BBT_CREATE |
>> NAND_BBT_WRITE
>>> + | NAND_BBT_2BIT | NAND_BBT_VERSION |
>> NAND_BBT_PERCHIP,
>>> + .offs = 4,
>>> + .len = 4,
>>> + .veroffs = 20,
>>> + .maxblocks = 4,
>>> + .pattern = bbt_pattern
>>> +};
>>> +
>>> +static struct nand_bbt_descr bbt_mirror_descr = {
>>> + .options = NAND_BBT_LASTBLOCK | NAND_BBT_CREATE |
>> NAND_BBT_WRITE
>>> + | NAND_BBT_2BIT | NAND_BBT_VERSION |
>> NAND_BBT_PERCHIP,
>>> + .offs = 4,
>>> + .len = 4,
>>> + .veroffs = 20,
>>> + .maxblocks = 4,
>>> + .pattern = mirror_pattern
>>> +};
>>> +
>>> +/**
>>> + * pl353_nand_read_buf_l - read chip data into buffer
>>> + * @chip: Pointer to the NAND chip info structure
>>> + * @in: Pointer to the buffer to store read data
>>> + * @len: Number of bytes to read
>>> + * Return: Always return zero
>>> + */
>>> +static int pl353_nand_read_buf_l(struct nand_chip *chip,
>>> + uint8_t *in,
>>> + unsigned int len)
>>> +{
>>> + int i;
>>> + unsigned long *ptr = (unsigned long *)in;
>>> +
>>> + len >>= 2;
>>
>> Can you please let the compiler optimize things? I don't find this very readable, I
>> would prefer a division here. And if this division by 4 is related to the size of *ptr,
>> please use the sizeof() macro. Otherwise please document this value.
> At a time, we are reading 4bytes. Hence >> 2.
> I didn't get your point.
> Are you saying instead of shifting, just use divide by 4?
>
>>
>>> + for (i = 0; i < len; i++)
>>> + ptr[i] = readl(chip->IO_ADDR_R);
>>
>> Space
> Ok, I will update it
>>
>>> + return 0;
>>> +}
>>> +
>>> +static void pl353_nand_write_buf_l(struct nand_chip *chip, const uint8_t
>> *buf,
>>> + int len)
>>> +{
>>> + int i;
>>> + unsigned long *ptr = (unsigned long *)buf;
>>> +
>>> + for (i = 0; i < len; i++)
>>> + writeb(ptr[i], chip->IO_ADDR_W);
>>
>> Here you use writeb (as opposed to readl previously). Then, I guess you can also
>> read byte per byte. If so, you can drop both helpers and let the core use its
>> defaults ones: nand_read/write_buf().
> May be the function name I have written wrongly.
> When using writel, it should be nand_write_buf_l.
> But the thing is, when using exec_op, core is not calling chip->read_byte(), hence I added
> Byte reading.
>>
>> Same for the next functions. Plus, if you don't use them inside
>> ->exec_op() implementation, they have to be removed anyway.
> The name of the function should change to buf_l, to do 4byte writes.
> The name is creating confusion.
>>
>>> +}
>>> +
>>> +/**
>>> + * pl353_nand_write_buf - write buffer to chip
>>> + * @mtd: Pointer to the mtd info structure
>>> + * @buf: Pointer to the buffer to store read data
>>> + * @len: Number of bytes to write
>>> + */
>>> +static void pl353_nand_write_buf(struct mtd_info *mtd, const uint8_t *buf,
>>> + int len)
>>> +{
>>> + int i;
>>> + struct nand_chip *chip = mtd_to_nand(mtd);
>>> + unsigned long *ptr = (unsigned long *)buf;
>>> +
>>> + len >>= 2;
>>> +
>>> + for (i = 0; i < len; i++)
>>> + writel(ptr[i], chip->IO_ADDR_W);
>>> +}
>>> +
>>> +/**
>>> + * pl353_nand_read_buf - read chip data into buffer
>>> + * @chip: Pointer to the NAND chip info structure
>>> + * @in: Pointer to the buffer to store read data
>>> + * @len: Number of bytes to read
>>> + * Return: 0 on success or error value on failure
>>> + */
>>> +static int pl353_nand_read_buf(struct nand_chip *chip,
>>> + uint8_t *in,
>>> + unsigned int len)
>>> +{
>>> + int i;
>>> +
>>> + for (i = 0; i < len; i++)
>>> + in[i] = readb(chip->IO_ADDR_R);
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +/**
>>> + * pl353_nand_calculate_hwecc - Calculate Hardware ECC
>>> + * @mtd: Pointer to the mtd_info structure
>>> + * @data: Pointer to the page data
>>> + * @ecc_code: Pointer to the ECC buffer where ECC data needs to be
>> stored
>>
>> You store ECC in a variable called "code", can you please make it consistent?
> Miquel, I am not using any variable called "code"
>>
>>> + *
>>> + * This function retrieves the Hardware ECC data from the controller
>>> +and returns
>>> + * ECC data back to the MTD subsystem.
>>> + *
>>> + * Return: 0 on success or error value on failure
>>> + */
>>> +static int pl353_nand_calculate_hwecc(struct mtd_info *mtd,
>>> + const u8 *data, u8 *ecc_code)
>>> +{
>>> + u32 ecc_value, ecc_status;
>>> + u8 ecc_reg, ecc_byte;
>>> + unsigned long timeout = jiffies + PL353_NAND_ECC_BUSY_TIMEOUT;
>>> + /* Wait till the ECC operation is complete or timeout */
>>> + do {
>>> + if (pl353_smc_ecc_is_busy())
>>
>> Where does this function come from?
> The pl353 SMC has memory controller driver and this NAND driver is using those APIs.
> I sent patches to add the memory controller driver for pl353.
> https://urldefense.proofpoint.com/v2/url?u=https-3A__www.spinics.net_lists_kernel_msg2748832.html&d=DwIGaQ&c=I_0YwoKy7z5LMTVdyO6YCiE2uzI1jjZZuIPelcSjixA&r=8Bziuw3IaCGjyrSAphuGwHmVdHcVwza-srUYwL9U_Ms&m=Qud5yLYSsZO8CluBVnAoEJUNWNJcqCra13xxjmbxyxs&s=8kxJPM5pXwV9aHPGTi534Vr_fwSjFLWJjopTzkQsSfk&e=
> https://urldefense.proofpoint.com/v2/url?u=https-3A__www.spinics.net_lists_kernel_msg2748834.html&d=DwIGaQ&c=I_0YwoKy7z5LMTVdyO6YCiE2uzI1jjZZuIPelcSjixA&r=8Bziuw3IaCGjyrSAphuGwHmVdHcVwza-srUYwL9U_Ms&m=Qud5yLYSsZO8CluBVnAoEJUNWNJcqCra13xxjmbxyxs&s=6RXTO4n9L2Frt2oHJ6H1QWvgq057zZWH2W3fsNrW6gc&e=
> https://urldefense.proofpoint.com/v2/url?u=https-3A__www.spinics.net_lists_kernel_msg2748840.html&d=DwIGaQ&c=I_0YwoKy7z5LMTVdyO6YCiE2uzI1jjZZuIPelcSjixA&r=8Bziuw3IaCGjyrSAphuGwHmVdHcVwza-srUYwL9U_Ms&m=Qud5yLYSsZO8CluBVnAoEJUNWNJcqCra13xxjmbxyxs&s=hCLzAMnGjHym1NwTglIEhxbypkoDyoHVvLHInmguBVo&e=
>
>>
>> I would rather prefer a readl_relaxed_poll_timeout() or similar, if possible.
> Let me check this function to use.
>>
>>> + cpu_relax();
>>> + else
>>> + break;
>>> + } while (!time_after_eq(jiffies, timeout));
>>> +
>>> + if (time_after_eq(jiffies, timeout)) {
>>> + pr_err("%s timed out\n", __func__);
>>> + return -ETIMEDOUT;
>>> + }
>>> +
>>> + for (ecc_reg = 0; ecc_reg < 4; ecc_reg++) {
>>> + /* Read ECC value for each block */
>>> + ecc_value = pl353_smc_get_ecc_val(ecc_reg);
>>
>> I don't have this function neither?
> Same as said above.
>>
>>> + ecc_status = (ecc_value >> 24) & 0xFF;
>>> + /* ECC value valid */
>>> + if (ecc_status & 0x40) {
>>> + for (ecc_byte = 0; ecc_byte < 3; ecc_byte++) {
>>> + /* Copy ECC bytes to MTD buffer */
>>> + *ecc_code = ~ecc_value & 0xFF;
>>> + ecc_value = ecc_value >> 8;
>>> + ecc_code++;
>>> + }
>>> + } else {
>>> + pr_warn("%s status failed\n", __func__);
>>> + return -1;
>>> + }
>>> + }
>>> + return 0;
>>> +}
>>> +
>>> +/**
>>> + * onehot - onehot function
>>> + * @value: Value to check for onehot
>>> + *
>>> + * This function checks whether a value is onehot or not.
>>> + * onehot is if and only if onebit is set.
>>
>> s/onebit/one bit/
>>
>> But I am sure there is already a function for that in the kernel, please don't use
>> your own implementation for this kind of stuff.
> Ok, I will check the existing function and will modify.
>>
>>> + *
>>> + * Return: 1 if it is onehot else 0
>>> + */
>>> +static int onehot(unsigned short value) {
>>> + return (value & (value - 1)) == 0;
>>> +}
>>> +
>>> +/**
>>> + * pl353_nand_correct_data - ECC correction function
>>> + * @mtd: Pointer to the mtd_info structure
>>> + * @buf: Pointer to the page data
>>> + * @read_ecc: Pointer to the ECC value read from spare data area
>>> + * @calc_ecc: Pointer to the calculated ECC value
>>> + *
>>> + * This function corrects the ECC single bit errors & detects 2-bit errors.
>>> + *
>>> + * Return: 0 if no ECC errors found
>>> + * 1 if single bit error found and corrected.
>>> + * -1 if multiple ECC errors found.
>>> + */
>>> +static int pl353_nand_correct_data(struct mtd_info *mtd, unsigned char
>> *buf,
>>> + unsigned char *read_ecc,
>>> + unsigned char *calc_ecc)
>>> +{
>>> + unsigned char bit_addr;
>>> + unsigned int byte_addr;
>>> + unsigned short ecc_odd, ecc_even, read_ecc_lower, read_ecc_upper;
>>> + unsigned short calc_ecc_lower, calc_ecc_upper;
>>> +
>>> + read_ecc_lower = (read_ecc[0] | (read_ecc[1] << 8)) & 0xfff;
>>> + read_ecc_upper = ((read_ecc[1] >> 4) | (read_ecc[2] << 4)) & 0xfff;
>>> +
>>> + calc_ecc_lower = (calc_ecc[0] | (calc_ecc[1] << 8)) & 0xfff;
>>> + calc_ecc_upper = ((calc_ecc[1] >> 4) | (calc_ecc[2] << 4)) & 0xfff;
>>> +
>>> + ecc_odd = read_ecc_lower ^ calc_ecc_lower;
>>> + ecc_even = read_ecc_upper ^ calc_ecc_upper;
>>> +
>>> + if ((ecc_odd == 0) && (ecc_even == 0))
>>> + return 0; /* no error */
>>
>> What about:
> I think, the below logic is doing that. if (ecc_odd == (~ecc_even & 0xfff))
>>
>> if (!ecc_odd && !ecc_even)
>> return 0;
>>
>>> +
>>> + if (ecc_odd == (~ecc_even & 0xfff)) {
>>> + /* bits [11:3] of error code is byte offset */
>>> + byte_addr = (ecc_odd >> 3) & 0x1ff;
>>> + /* bits [2:0] of error code is bit offset */
>>> + bit_addr = ecc_odd & 0x7;
>>> + /* Toggling error bit */
>>> + buf[byte_addr] ^= (1 << bit_addr);
>>
>> Use BIT(bit_addr) macro instead?
> Ok, I will modify it.
>>
>>> + return 1;
>>> + }
>>> +
>>> + if (onehot(ecc_odd | ecc_even) == 1)
>>> + return 1; /* one error in parity */
>>
>> Comment should be before the if statement here.
> Ok, I will update it.
>>
>>> +
>>> + return -1; /* Uncorrectable error */
>>
>> And here, before the return statement.
> Ok, I will update it.
>>
>>> +}
>>> +
>>> +static int pl353_dev_timeout(struct mtd_info *mtd, struct nand_chip
>>> +*chip) {
>>> + unsigned long timeout = jiffies + PL353_NAND_DEV_BUSY_TIMEOUT;
>>> +
>>> + do {
>>> + if (chip->dev_ready(mtd))
>>> + break;
>>> + cpu_relax();
>>> + } while (!time_after_eq(jiffies, timeout));
>>> +
>>> + if (time_after_eq(jiffies, timeout)) {
>>> + pr_err("%s timed out\n", __func__);
>>> + return -1;
>>> + }
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +static void pl353_prepare_cmd(struct mtd_info *mtd, struct nand_chip *chip,
>>> + int page, int column, int start_cmd, int end_cmd, bool read) {
>>> + unsigned long data_phase_addr;
>>> + u32 end_cmd_valid = 0;
>>> + void __iomem *cmd_addr;
>>> + unsigned long cmd_phase_addr = 0, cmd_data = 0;
>>> +
>>> + struct pl353_nand_info *xnand =
>>> + container_of(chip, struct pl353_nand_info, chip);
>>> +
>>> + if (read)
>>> + end_cmd_valid = 1;
>>> + else
>>> + end_cmd_valid = 0;
>>
>> This is a good spot to use the ternary operator :)
> Yes, we can use. I will add this.
>>
>>> +
>>> + cmd_phase_addr = (unsigned long __force)xnand->nand_base + (
>>> + (((xnand->row_addr_cycles) + (xnand-
>>> col_addr_cycles))
>>> + << ADDR_CYCLES_SHIFT) |
>>> + (end_cmd_valid << END_CMD_VALID_SHIFT)
>> |
>>> + (COMMAND_PHASE) |
>>> + (end_cmd << END_CMD_SHIFT) |
>>
>> Please don't align the '|'
> You mean, tabbing?
>>
>>> + (start_cmd << START_CMD_SHIFT));
>>> + cmd_addr = (void __iomem * __force)cmd_phase_addr;
>>> +
>>> + /* Get the data phase address */
>>> + data_phase_addr = (unsigned long __force)xnand->nand_base + (
>>> + (0x0 << CLEAR_CS_SHIFT) |
>>> + (0 << END_CMD_VALID_SHIFT) |
>>> + (DATA_PHASE) |
>>> + (end_cmd << END_CMD_SHIFT) |
>>> + (0x0 << ECC_LAST_SHIFT));
>>> +
>>> + chip->IO_ADDR_R = (void __iomem * __force)data_phase_addr;
>>> + chip->IO_ADDR_W = chip->IO_ADDR_R;
>>> + if (chip->options & NAND_BUSWIDTH_16)
>>> + column >>= 1;
>>
>> / 2
>>
>>> + cmd_data = column;
>>> + if (mtd->writesize > PL353_NAND_ECC_SIZE) {
>>> + cmd_data |= page << 16;
>>> + /* Another address cycle for devices > 128MiB */
>>> + if (chip->chipsize > (128 << 20)) {
>>
>> Now there is a flag for that in the core, called NAND_ROW_ADDR_3.
> I will check and update.
>>
>>> + pl353_nand_write32(cmd_addr, cmd_data);
>>> + cmd_data = (page >> 16);
>>> + }
>>> + } else {
>>> + cmd_data |= page << 8;
>>> + }
>>
>> Space
> Ok, I will update.
>>
>>> + pl353_nand_write32(cmd_addr, cmd_data); }
>>> +
>>> +/**
>>> + * pl353_nand_read_oob - [REPLACEABLE] the most common OOB data read
>> function
>>> + * @mtd: Pointer to the mtd info structure
>>> + * @chip: Pointer to the NAND chip info structure
>>> + * @page: Page number to read
>>> + *
>>> + * Return: Always return zero
>>> + */
>>> +static int pl353_nand_read_oob(struct mtd_info *mtd, struct nand_chip
>> *chip,
>>> + int page)
>>> +{
>>> +
>>> + unsigned long data_phase_addr;
>>> + uint8_t *p;
>>> +
>>> + chip->pagebuf = -1;
>>> + if (mtd->writesize < PL353_NAND_ECC_SIZE)
>>> + return 0;
>>> +
>>> + pl353_prepare_cmd(mtd, chip, page, mtd->writesize,
>> NAND_CMD_READ0,
>>> + NAND_CMD_READSTART, 1);
>>
>> Alignment
> Are you running any script apart from checkpatch?
> Any way I will correct it.
>>
>>> +
>>> + ndelay(100);
>>> + pl353_dev_timeout(mtd, chip);
>>> +
>>> + p = chip->oob_poi;
>>> + pl353_nand_read_buf_l(chip, p,
>>> + (mtd->oobsize - PL353_NAND_LAST_TRANSFER_LENGTH));
>>
>> Alignment
> Ok, I will correct it.
>>
>>> + p += (mtd->oobsize - PL353_NAND_LAST_TRANSFER_LENGTH);
>>> +
>>> + data_phase_addr = (unsigned long __force)chip->IO_ADDR_R;
>>> + data_phase_addr |= PL353_NAND_CLEAR_CS;
>>> + chip->IO_ADDR_R = (void __iomem * __force)data_phase_addr;
>>> + pl353_nand_read_buf_l(chip, p,
>> PL353_NAND_LAST_TRANSFER_LENGTH);
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +/**
>>> + * pl353_nand_write_oob - [REPLACEABLE] the most common OOB data
>> write function
>>> + * @mtd: Pointer to the mtd info structure
>>> + * @chip: Pointer to the NAND chip info structure
>>> + * @page: Page number to write
>>> + *
>>> + * Return: Zero on success and EIO on failure
>>> + */
>>
>> Comments inside this function would be welcome!
> Ok, I will add.
>>
>>> +static int pl353_nand_write_oob(struct mtd_info *mtd, struct nand_chip
>> *chip,
>>> + int page)
>>> +{
>>> +
>>
>> Extra space
> Any way I will correct it.
>>
>>> + const uint8_t *buf = chip->oob_poi;
>>> + unsigned long data_phase_addr;
>>> + struct pl353_nand_info *xnand =
>>> + container_of(chip, struct pl353_nand_info, chip);
>>> + u32 addrcycles = 0, ret;
>>> + unsigned long timeout = jiffies + PL353_NAND_DEV_BUSY_TIMEOUT;
>>> + u8 status;
>>> +
>>> + chip->pagebuf = -1;
>>> + addrcycles = xnand->row_addr_cycles + xnand->col_addr_cycles;
>>> + pl353_prepare_cmd(mtd, chip, page, mtd->writesize,
>> NAND_CMD_SEQIN,
>>> + NAND_CMD_PAGEPROG, 0);
>>> + ndelay(100);
>>> + pl353_nand_write_buf(mtd, buf,
>>> + (mtd->oobsize -
>> PL353_NAND_LAST_TRANSFER_LENGTH));
>>> + buf += (mtd->oobsize - PL353_NAND_LAST_TRANSFER_LENGTH);
>>> +
>>> + data_phase_addr = (unsigned long __force)chip->IO_ADDR_W;
>>> + data_phase_addr |= PL353_NAND_CLEAR_CS;
>>> + data_phase_addr |= (1 << END_CMD_VALID_SHIFT);
>>> + chip->IO_ADDR_W = (void __iomem * __force)data_phase_addr;
>>> + pl353_nand_write_buf(mtd, buf,
>> PL353_NAND_LAST_TRANSFER_LENGTH);
>>> +
>>> + /* Send command to program the OOB data */
>>> + ret = nand_status_op(chip, &status);
>>
>> Doing this without exiting the status state is broken.
> I didn't get.
>>
>>> + timeout = jiffies + msecs_to_jiffies(timeout);
>>> + do {
>>> + if (chip->dev_ready) {
>>> + if (chip->dev_ready(mtd))
>>> + break;
>>> + } else {
>>
>> You give your own implementation of ->dev_ready(). So this is dead code.
> Hmm, you are correct, I will remove this code.
>>
>>> + if (status & NAND_STATUS_READY)
>>
>> You don't update "status", while you wait for it to change.
> Basically I am waiting for device ready, but not updating the status.
> I will look this implementation and will get back to you.
>>
>>> + break;
>>> + }
>>> + cond_resched();
>>> + } while (time_before(jiffies, timeout));
>>> +
>>> + /* This can happen if in case of timeout or buggy dev_ready */
>>> + WARN_ON(!(status & NAND_STATUS_READY));
>>
>> I think the whole block has to be replaced by a simple nand_wait_ready() call.
> Let me use this, thanks for pointing.
>>
>>> +
>>> + return (status & NAND_STATUS_FAIL) ? -EIO : 0; }
>>> +
>>> +/**
>>> + * pl353_nand_read_page_raw - [Intern] read raw page data without ecc
>>> + * @mtd: Pointer to the mtd info structure
>>> + * @chip: Pointer to the NAND chip info structure
>>> + * @buf: Pointer to the data buffer
>>> + * @oob_required: Caller requires OOB data read to chip->oob_poi
>>> + * @page: Page number to read
>>> + *
>>> + * Return: Always return zero
>>> + */
>>> +static int pl353_nand_read_page_raw(struct mtd_info *mtd,
>>> + struct nand_chip *chip,
>>> + uint8_t *buf, int oob_required, int page)
>>
>> Do you really need raw accessors?
> Yes, when using on-die ecc, this function is getting called.
> i.e. nand_micron.c is calling nand_set_features_op, with DATA_OUT_INSTR, and there
> we are using this.
>>
>> Not sure this is needed.
>>
>>> +{
>>> + unsigned long data_phase_addr;
>>> + uint8_t *p;
>>> +
>>> + pl353_nand_read_buf_l(chip, buf, mtd->writesize);
>>> + p = chip->oob_poi;
>>> + pl353_nand_read_buf_l(chip, p,
>>> + (mtd->oobsize - PL353_NAND_LAST_TRANSFER_LENGTH));
>>> + p += (mtd->oobsize - PL353_NAND_LAST_TRANSFER_LENGTH);
>>> +
>>> + data_phase_addr = (unsigned long __force)chip->IO_ADDR_R;
>>> + data_phase_addr |= PL353_NAND_CLEAR_CS;
>>> + chip->IO_ADDR_R = (void __iomem * __force)data_phase_addr;
>>> +
>>> + pl353_nand_read_buf_l(chip, p,
>> PL353_NAND_LAST_TRANSFER_LENGTH);
>>> + return 0;
>>> +}
>>> +
>>> +/**
>>> + * pl353_nand_write_page_raw - [Intern] raw page write function
>>> + * @mtd: Pointer to the mtd info structure
>>> + * @chip: Pointer to the NAND chip info structure
>>> + * @buf: Pointer to the data buffer
>>> + * @oob_required: Caller requires OOB data read to chip->oob_poi
>>> + * @page: Page number to write
>>> + *
>>> + * Return: Always return zero
>>> + */
>>> +static int pl353_nand_write_page_raw(struct mtd_info *mtd,
>>> + struct nand_chip *chip,
>>> + const uint8_t *buf, int oob_required,
>>> + int page)
>>> +{
>>> + unsigned long data_phase_addr;
>>> + uint8_t *p;
>>> +
>>> + pl353_prepare_cmd(mtd, chip, page, 0, NAND_CMD_SEQIN,
>>> + NAND_CMD_PAGEPROG, 0);
>>> + pl353_nand_write_buf(mtd, buf, mtd->writesize);
>>> + p = chip->oob_poi;
>>> + pl353_nand_write_buf(mtd, p,
>>> + (mtd->oobsize -
>> PL353_NAND_LAST_TRANSFER_LENGTH));
>>> + p += (mtd->oobsize - PL353_NAND_LAST_TRANSFER_LENGTH);
>>> +
>>> + data_phase_addr = (unsigned long __force)chip->IO_ADDR_W;
>>> + data_phase_addr |= PL353_NAND_CLEAR_CS;
>>> + data_phase_addr |= (1 << END_CMD_VALID_SHIFT);
>>> + chip->IO_ADDR_W = (void __iomem * __force)data_phase_addr;
>>> +
>>> + pl353_nand_write_buf(mtd, p,
>> PL353_NAND_LAST_TRANSFER_LENGTH);
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +/**
>>> + * nand_write_page_hwecc - Hardware ECC based page write function
>>> + * @mtd: Pointer to the mtd info structure
>>> + * @chip: Pointer to the NAND chip info structure
>>> + * @buf: Pointer to the data buffer
>>> + * @oob_required: Caller requires OOB data read to chip->oob_poi
>>> + * @page: Page number to write
>>> + *
>>> + * This functions writes data and hardware generated ECC values in to the
>> page.
>>> + *
>>> + * Return: Always return zero
>>> + */
>>> +static int pl353_nand_write_page_hwecc(struct mtd_info *mtd,
>>> + struct nand_chip *chip, const uint8_t *buf,
>>> + int oob_required, int page)
>>> +{
>>> +
>>> + int eccsize = chip->ecc.size;
>>> + int eccsteps = chip->ecc.steps;
>>> + uint8_t *ecc_calc = chip->ecc.calc_buf;
>>> + const uint8_t *p = buf;
>>> + uint8_t *oob_ptr;
>>> + u32 ret;
>>> + unsigned long data_phase_addr, timeo;
>>> + u8 status;
>>> +
>>> + pl353_prepare_cmd(mtd, chip, page, 0, NAND_CMD_SEQIN,
>>> + NAND_CMD_PAGEPROG, 0);
>>> + ndelay(100);
>>> + for ( ; (eccsteps - 1); eccsteps--) {
>>> + pl353_nand_write_buf(mtd, p, eccsize);
>>> + p += eccsize;
>>> + }
>>> + pl353_nand_write_buf(mtd, p,
>>> + (eccsize - PL353_NAND_LAST_TRANSFER_LENGTH));
>>> + p += (eccsize - PL353_NAND_LAST_TRANSFER_LENGTH);
>>> +
>>> + /* Set ECC Last bit to 1 */
>>> + data_phase_addr = (unsigned long __force)chip->IO_ADDR_W;
>>> + data_phase_addr |= PL353_NAND_ECC_LAST;
>>> + chip->IO_ADDR_W = (void __iomem * __force)data_phase_addr;
>>> + pl353_nand_write_buf(mtd, p,
>> PL353_NAND_LAST_TRANSFER_LENGTH);
>>> +
>>> + p = buf;
>>> + chip->ecc.calculate(mtd, p, &ecc_calc[0]);
>>> +
>>> + /* Wait for ECC to be calculated and read the error values */
>>> + ret = mtd_ooblayout_set_eccbytes(mtd, ecc_calc, chip->oob_poi,
>>> + 0, chip->ecc.total);
>>> + if (ret)
>>> + return ret;
>>> + /* Clear ECC last bit */
>>> + data_phase_addr = (unsigned long __force)chip->IO_ADDR_W;
>>> + data_phase_addr &= ~PL353_NAND_ECC_LAST;
>>> + chip->IO_ADDR_W = (void __iomem * __force)data_phase_addr;
>>> +
>>> + /* Write the spare area with ECC bytes */
>>> + oob_ptr = chip->oob_poi;
>>> + pl353_nand_write_buf(mtd, oob_ptr,
>>> + (mtd->oobsize -
>> PL353_NAND_LAST_TRANSFER_LENGTH));
>>> +
>>> + data_phase_addr = (unsigned long __force)chip->IO_ADDR_W;
>>> + data_phase_addr |= PL353_NAND_CLEAR_CS;
>>> + data_phase_addr |= (1 << END_CMD_VALID_SHIFT);
>>> + chip->IO_ADDR_W = (void __iomem * __force)data_phase_addr;
>>> + oob_ptr += (mtd->oobsize - PL353_NAND_LAST_TRANSFER_LENGTH);
>>> + pl353_nand_write_buf(mtd, oob_ptr,
>> PL353_NAND_LAST_TRANSFER_LENGTH);
>>
>> Once using ->exec_op(), you should use something like nand_write_data_op().
> Yes, I will update this.
>>> +
>>> + /*
>>> + * Apply this short delay always to ensure that we do wait tWB in any
>>> + * case on any machine.
>>> + */
>>> + ndelay(100);
>>> + ret = nand_status_op(chip, &status);
>>> + timeo = jiffies + msecs_to_jiffies(400);
>>> + do {
>>> + if (chip->dev_ready) {
>>> + if (chip->dev_ready(mtd))
>>> + break;
>>> + } else {
>>> + if (status & NAND_STATUS_READY)
>>> + break;
>>> + }
>>> + cond_resched();
>>> + } while (time_before(jiffies, timeo));
>>> +
>>> + /* This can happen if in case of timeout or buggy dev_ready */
>>> + WARN_ON(!(status & NAND_STATUS_READY));
>>> +
>>
>> Same here about ->dev_ready().
> Ok
>>
>>> + return (status & NAND_STATUS_FAIL) ? -EIO : 0; }
>>> +
>>> +/**
>>> + * pl353_nand_write_page_swecc - [REPLACEABLE] software ecc based page
>> write
>>> + * function
>>> + * @mtd: Pointer to the mtd info structure
>>> + * @chip: Pointer to the NAND chip info structure
>>> + * @buf: Pointer to the data buffer
>>> + * @oob_required: Caller requires OOB data read to chip->oob_poi
>>> + * @page: Page number to write
>>> + *
>>> + * Return: Always return zero
>>> + */
>>> +static int pl353_nand_write_page_swecc(struct mtd_info *mtd,
>>> + struct nand_chip *chip, const uint8_t *buf,
>>> + int oob_required, int page)
>>> +{
>>
>> This is an exact copy of the core's function, you probably don't need it.
> Ok, I will remove it in next version.
>>
>>> + int i, eccsize = chip->ecc.size;
>>> + int eccbytes = chip->ecc.bytes;
>>> + int eccsteps = chip->ecc.steps;
>>> + uint8_t *ecc_calc = chip->ecc.calc_buf;
>>> + const uint8_t *p = buf;
>>> + u32 ret;
>>> +
>>> + for (i = 0; eccsteps; eccsteps--, i += eccbytes, p += eccsize)
>>> + chip->ecc.calculate(mtd, p, &ecc_calc[0]);
>>> +
>>> + ret = mtd_ooblayout_set_eccbytes(mtd, ecc_calc, chip->oob_poi,
>>> + 0, chip->ecc.total);
>>> + if (ret)
>>> + return ret;
>>> + chip->ecc.write_page_raw(mtd, chip, buf, 1, page);
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +/**
>>> + * pl353_nand_read_page_hwecc - Hardware ECC based page read function
>>> + * @mtd: Pointer to the mtd info structure
>>> + * @chip: Pointer to the NAND chip info structure
>>> + * @buf: Pointer to the buffer to store read data
>>> + * @oob_required: Caller requires OOB data read to chip->oob_poi
>>> + * @page: Page number to read
>>> + *
>>> + * This functions reads data and checks the data integrity by
>>> +comparing hardware
>>> + * generated ECC values and read ECC values from spare area.
>>> + *
>>> + * Return: 0 always and updates ECC operation status in to MTD structure
>>> + */
>>> +static int pl353_nand_read_page_hwecc(struct mtd_info *mtd,
>>> + struct nand_chip *chip,
>>> + uint8_t *buf, int oob_required, int page) {
>>> + int i, stat, eccsize = chip->ecc.size;
>>> + int eccbytes = chip->ecc.bytes;
>>> + int eccsteps = chip->ecc.steps;
>>> + uint8_t *p = buf;
>>> + uint8_t *ecc_calc = chip->ecc.calc_buf;
>>> + uint8_t *ecc_code = chip->ecc.code_buf;
>>> +
>>> + uint8_t *oob_ptr;
>>> + u32 ret;
>>> + unsigned long data_phase_addr;
>>> + unsigned long timeout = jiffies + PL353_NAND_DEV_BUSY_TIMEOUT;
>>> +
>>> + pl353_prepare_cmd(mtd, chip, page, 0, NAND_CMD_READ0,
>>> + NAND_CMD_READSTART, 1);
>>> + ndelay(100);
>>> + do {
>>> + if (chip->dev_ready(mtd))
>>> + break;
>>> + cpu_relax();
>>> + } while (!time_after_eq(jiffies, timeout));
>>> +
>>> + if (time_after_eq(jiffies, timeout)) {
>>> + pr_err("%s timed out\n", __func__);
>>> + return -1;
>>> + }
>>> + for ( ; (eccsteps - 1); eccsteps--) {
>>> + pl353_nand_read_buf_l(chip, p, eccsize);
>>> + p += eccsize;
>>> + }
>>> + pl353_nand_read_buf_l(chip, p,
>>> + (eccsize - PL353_NAND_LAST_TRANSFER_LENGTH));
>>> + p += (eccsize - PL353_NAND_LAST_TRANSFER_LENGTH);
>>> +
>>> + /* Set ECC Last bit to 1 */
>>> + data_phase_addr = (unsigned long __force)chip->IO_ADDR_R;
>>> + data_phase_addr |= PL353_NAND_ECC_LAST;
>>> + chip->IO_ADDR_R = (void __iomem * __force)data_phase_addr;
>>> + pl353_nand_read_buf_l(chip, p,
>> PL353_NAND_LAST_TRANSFER_LENGTH);
>>> +
>>> + /* Read the calculated ECC value */
>>> + p = buf;
>>> + chip->ecc.calculate(mtd, p, &ecc_calc[0]);
>>> +
>>> + /* Clear ECC last bit */
>>> + data_phase_addr = (unsigned long __force)chip->IO_ADDR_R;
>>> + data_phase_addr &= ~PL353_NAND_ECC_LAST;
>>> + chip->IO_ADDR_R = (void __iomem * __force)data_phase_addr;
>>> +
>>> + /* Read the stored ECC value */
>>> + oob_ptr = chip->oob_poi;
>>> + pl353_nand_read_buf_l(chip, oob_ptr,
>>> + (mtd->oobsize - PL353_NAND_LAST_TRANSFER_LENGTH));
>>> +
>>> + /* de-assert chip select */
>>> + data_phase_addr = (unsigned long __force)chip->IO_ADDR_R;
>>> + data_phase_addr |= PL353_NAND_CLEAR_CS;
>>> + chip->IO_ADDR_R = (void __iomem * __force)data_phase_addr;
>>> +
>>> + oob_ptr += (mtd->oobsize - PL353_NAND_LAST_TRANSFER_LENGTH);
>>> + pl353_nand_read_buf_l(chip, oob_ptr,
>>> +PL353_NAND_LAST_TRANSFER_LENGTH);
>>> +
>>> + ret = mtd_ooblayout_get_eccbytes(mtd, ecc_code, chip->oob_poi, 0,
>>> + chip->ecc.total);
>>> + if (ret)
>>> + return ret;
>>> +
>>> + eccsteps = chip->ecc.steps;
>>> + p = buf;
>>> +
>>> + /* Check ECC error for all blocks and correct if it is correctable */
>>> + for (i = 0 ; eccsteps; eccsteps--, i += eccbytes, p += eccsize) {
>>
>> In this loop you should monitore a max_bitflips value and return it instead of
>> zero at the end of the function.
> Ok, I will add it.
>>
>>> + stat = chip->ecc.correct(mtd, p, &ecc_code[i], &ecc_calc[i]);
>>> + if (stat < 0)
>>> + mtd->ecc_stats.failed++;
>>> + else
>>> + mtd->ecc_stats.corrected += stat;
>>> + }
>>> + return 0;
>>> +}
>>> +
>>> +/**
>>> + * pl353_nand_read_page_swecc - [REPLACEABLE] software ecc based page
>> read
>>> + * function
>>> + * @mtd: Pointer to the mtd info structure
>>> + * @chip: Pointer to the NAND chip info structure
>>> + * @buf: Pointer to the buffer to store read data
>>> + * @oob_required: Caller requires OOB data read to chip->oob_poi
>>> + * @page: Page number to read
>>> + *
>>> + * Return: Always return zero
>>> + */
>>> +static int pl353_nand_read_page_swecc(struct mtd_info *mtd,
>>> + struct nand_chip *chip,
>>> + uint8_t *buf, int oob_required, int page) {
>>
>> Same, you probably don't need this function.
> Yes, I will remove in next version.
>>
>>> + int i, eccsize = chip->ecc.size;
>>> + int eccbytes = chip->ecc.bytes;
>>> + int eccsteps = chip->ecc.steps;
>>> + uint8_t *p = buf;
>>> + uint8_t *ecc_calc = chip->ecc.calc_buf;
>>> + uint8_t *ecc_code = chip->ecc.code_buf;
>>> + u32 ret;
>>> +
>>> + chip->ecc.read_page_raw(mtd, chip, buf, page, 1);
>>> +
>>> + for (i = 0; eccsteps; eccsteps--, i += eccbytes, p += eccsize)
>>> + chip->ecc.calculate(mtd, p, &ecc_calc[i]);
>>> +
>>> + ret = mtd_ooblayout_get_eccbytes(mtd, ecc_calc, chip->oob_poi,
>>> + 0, chip->ecc.total);
>>> +
>>> + eccsteps = chip->ecc.steps;
>>> + p = buf;
>>> +
>>> + for (i = 0 ; eccsteps; eccsteps--, i += eccbytes, p += eccsize) {
>>> + int stat;
>>> +
>>> + stat = chip->ecc.correct(mtd, p, &ecc_code[i], &ecc_calc[i]);
>>> + if (stat < 0)
>>> + mtd->ecc_stats.failed++;
>>> + else
>>> + mtd->ecc_stats.corrected += stat;
>>> + }
>>> + return 0;
>>> +}
>>> +
>>> +/**
>>> + * pl353_nand_select_chip - Select the flash device
>>> + * @mtd: Pointer to the mtd info structure
>>> + * @chip: Pointer to the NAND chip info structure
>>> + *
>>> + * This function is empty as the NAND controller handles chip select
>>> +line
>>> + * internally based on the chip address passed in command and data phase.
>>> + */
>>> +static void pl353_nand_select_chip(struct mtd_info *mtd, int chip) {
>>> +}
>>> +
>>> +/* NAND framework ->exec_op() hooks and related helpers */ static
>>> +void pl353_nfc_parse_instructions(struct nand_chip *chip,
>>> + const struct nand_subop *subop,
>>> + struct pl353_nfc_op *nfc_op)
>>> +{
>>> + const struct nand_op_instr *instr = NULL;
>>> + unsigned int op_id, offset, naddrs;
>>> + int i;
>>> + const u8 *addrs;
>>> +
>>> + memset(nfc_op, 0, sizeof(struct pl353_nfc_op));
>>> + for (op_id = 0; op_id < subop->ninstrs; op_id++) {
>>> +
>>
>> What is this for-loop for? I don't get it as you break the switch in every case?
> I think, breaking switch case only not for loop.
>>
>>> + nfc_op->len = nand_subop_get_data_len(subop, op_id);
>>> +
>>> + instr = &subop->instrs[op_id];
>>> + if (subop->ninstrs == 1)
>>> + nfc_op->cmnds[0] = -1;
>>> + switch (instr->type) {
>>> + case NAND_OP_CMD_INSTR:
>>> + nfc_op->type = NAND_OP_CMD_INSTR;
>>> + nfc_op->end_cmd = op_id - 1;
>>> + if (op_id)
>>
>> You should put { } on the if also if the else statement needs braces.
> Ok, but I didn't see any warning from checkpatch.
>>
>>> + nfc_op->cmnds[1] = instr->ctx.cmd.opcode;
>>> + else {
>>> + nfc_op->cmnds[0] = instr->ctx.cmd.opcode;
>>> + nfc_op->cmnds[1] = -1;
>>> + }
>>> + break;
>>> +
>>> + case NAND_OP_ADDR_INSTR:
>>> + offset = nand_subop_get_addr_start_off(subop, op_id);
>>> + naddrs = nand_subop_get_num_addr_cyc(subop,
>> op_id);
>>> + addrs = &instr->ctx.addr.addrs[offset];
>>> + nfc_op->addrs = instr->ctx.addr.addrs[offset];
>>> + for (i = 0; i < min_t(unsigned int, 4, naddrs); i++)
>>> + nfc_op->addrs |= instr->ctx.addr.addrs[i] <<
>>> + (8 * i);
>>> +
>>> + if (naddrs >= 5) {
>>> + nfc_op->addrs >>= 16;
>>> + nfc_op->addrs |= (addrs[4] << 16);
>>> + nfc_op->thirdrow = 1;
>>> + }
>>> + nfc_op->naddrs = nand_subop_get_num_addr_cyc
>>> + (subop, op_id);
>>
>> Don't put the parameters of a function on the next line like that?
> Ok, I will update it.
>>
>>> + break;
>>> +
>>> + case NAND_OP_DATA_IN_INSTR:
>>> + nfc_op->data_instr = instr;
>>> + nfc_op->type = NAND_OP_DATA_IN_INSTR;
>>> + nfc_op->data_instr_idx = op_id;
>>> + break;
>>> +
>>> + case NAND_OP_DATA_OUT_INSTR:
>>> + nfc_op->data_instr = instr;
>>> + nfc_op->type = NAND_OP_DATA_IN_INSTR;
>> DATA_OUT_INSTR?
> Yes, I will correct it.
> But I am using instr->type in cmd_function.
> I will update it.
>>
>> Is this really tested?
> Yes, During on-die ECC test.
> When nand_micron is asking to set some features, this will execute.
>>
>>> + nfc_op->data_instr_idx = op_id;
>>> + break;
>>> +
>>> + case NAND_OP_WAITRDY_INSTR:
>>> + nfc_op->rdy_timeout_ms = instr-
>>> ctx.waitrdy.timeout_ms;
>>> + nfc_op->rdy_delay_ns = instr->delay_ns;
>>> + nfc_op->wait = true;
>>> + break;
>>> + }
>>> + }
>>> +}
>>> +
>>> +/**
>>> + * pl353_nand_cmd_function - Send command to NAND device
>>> + * @chip: Pointer to the NAND chip info structure
>>> + * @subop: Pointer to array of instructions
>>> + * Return: Always return zero
>>> + */
>>> +static int pl353_nand_cmd_function(struct nand_chip *chip,
>>> + const struct nand_subop *subop) {
>>> + struct mtd_info *mtd = nand_to_mtd(chip);
>>> + const struct nand_op_instr *instr;
>>> + struct pl353_nfc_op nfc_op;
>>
>> nfc_op = {}; to initialize to 0.
> Ok, I will update it.
>>
>>> + struct pl353_nand_info *xnand =
>>> + container_of(chip, struct pl353_nand_info, chip);
>>> + void __iomem *cmd_addr;
>>> + unsigned long cmd_data = 0, end_cmd_valid = 0;
>>> + unsigned long cmd_phase_addr, data_phase_addr, end_cmd;
>>> + unsigned long timeout = jiffies + PL353_NAND_DEV_BUSY_TIMEOUT;
>>> + u32 addrcycles = 0;
>>> + unsigned int op_id, len, offset;
>>> +
>>> + pl353_nfc_parse_instructions(chip, subop, &nfc_op);
>>> + instr = nfc_op.data_instr;
>>> + op_id = nfc_op.data_instr_idx;
>>> + len = nand_subop_get_data_len(subop, op_id);
>>> + offset = nand_subop_get_data_start_off(subop, op_id);
>>> +
>>> + if (nfc_op.cmnds[0] != -1) {
>>> + if (xnand->end_cmd_pending) {
>>> + /*
>>> + * Check for end command if this command request is
>>> + * same as the pending command then return
>>> + */
>>> + if (xnand->end_cmd == nfc_op.cmnds[0]) {
>>> + xnand->end_cmd = 0;
>>> + xnand->end_cmd_pending = 0;
>>> + return 0;
>>> + }
>>> + }
>>> +
>>> + /* Clear interrupt */
>>> + pl353_smc_clr_nand_int();
>>> + end_cmd_valid = 0;
>>
>> Space
> Ok, I will correct it.
>>
>>> + /* Get the command phase address */
>>> + if (nfc_op.cmnds[1] != -1) {
>>> + end_cmd_valid = 1;
>>> + } else {
>>> + if (nfc_op.cmnds[0] == NAND_CMD_READ0)
>>> + return 0;
>>> + }
>>> + if (nfc_op.end_cmd == NAND_CMD_NONE)
>>> + end_cmd = 0x0;
>>> + else
>>> + end_cmd = nfc_op.cmnds[1];
>>> +
>>> + addrcycles = nfc_op.naddrs;
>>> + if (nfc_op.cmnds[0] == NAND_CMD_READ0 ||
>>> + nfc_op.cmnds[0] == NAND_CMD_SEQIN)
>>> + addrcycles = xnand->row_addr_cycles +
>>> + xnand->col_addr_cycles;
>>> + else if ((nfc_op.cmnds[0] == NAND_CMD_ERASE1) ||
>>> + (nfc_op.cmnds[0] == NAND_CMD_ERASE2))
>>> + addrcycles = xnand->row_addr_cycles;
>>> + else
>>> + addrcycles = nfc_op.naddrs;
>>
>> A switch block would probably be appropriate.
> Yes we can use.
>>
>>> + cmd_phase_addr = (unsigned long __force)xnand->nand_base +
>> (
>>> + (addrcycles << ADDR_CYCLES_SHIFT) |
>>> + (end_cmd_valid << END_CMD_VALID_SHIFT)
>> |
>>> + (COMMAND_PHASE) |
>>> + (end_cmd << END_CMD_SHIFT) |
>>> + (nfc_op.cmnds[0] << START_CMD_SHIFT));
>>> +
>>> + cmd_addr = (void __iomem * __force)cmd_phase_addr;
>>
>> Space
> Ok, I will correc it.
>>
>>> + /* Get the data phase address */
>>> + end_cmd_valid = 0;
>>> +
>>> + data_phase_addr = (unsigned long __force)xnand->nand_base +
>> (
>>> + (0x0 << CLEAR_CS_SHIFT) |
>>> + (end_cmd_valid << END_CMD_VALID_SHIFT)|
>>> + (DATA_PHASE) |
>>> + (end_cmd << END_CMD_SHIFT) |
>>> + (0x0 << ECC_LAST_SHIFT));
>>> + chip->IO_ADDR_R = (void __iomem *
>> __force)data_phase_addr;
>>
>> Do you really need this "__force" ?
> Let me check and get back to you on this.
>>
>>> + chip->IO_ADDR_W = chip->IO_ADDR_R;
>>> + /* Command phase AXI write */
>>> + /* Read & Write */
>>> + if (nfc_op.thirdrow) {
>>> + nfc_op.thirdrow = 0;
>>> + if (mtd->writesize > PL353_NAND_ECC_SIZE) {
>>> + cmd_data |= nfc_op.addrs << 16;
>>> + /* Another address cycle for devices > 128MiB
>> */
>>> + if (chip->chipsize > (128 << 20)) {
>>> + pl353_nand_write32(cmd_addr,
>> cmd_data);
>>> + cmd_data = (nfc_op.addrs >> 16);
>>> + }
>>> + }
>>> + } else {
>>> + if (nfc_op.addrs != -1) {
>>> + int column = nfc_op.addrs;
>>> + /*
>>> + * Change read/write column, read id etc
>>> + * Adjust columns for 16 bit bus width
>>> + */
>>> + if ((chip->options & NAND_BUSWIDTH_16) &&
>>> + ((nfc_op.cmnds[0] == NAND_CMD_READ0) ||
>>> + (nfc_op.cmnds[0] == NAND_CMD_SEQIN) ||
>>> + (nfc_op.cmnds[0] == NAND_CMD_RNDOUT) ||
>>> + (nfc_op.cmnds[0] == NAND_CMD_RNDIN))) {
>>
>> Alignment.
> Ok, I will update it.
>>
>>> + column >>= 1;
>>> + }
>>> + cmd_data = nfc_op.addrs;
>>> + }
>>> + }
>>> + pl353_nand_write32(cmd_addr, cmd_data);
>>> + if (nfc_op.type != 0) {
>>> + xnand->end_cmd = nfc_op.end_cmd;
>>> + xnand->end_cmd_pending = 1;
>>> + }
>>> + ndelay(100);
>>
>> Why?
> I found some errors during on_die testing, hence I added this delay.
> Let me check once

Either the controller or nand itself requires a settle period after
commands are delivered, before data may be read/written. As I
understand, this ndelay() is intended to facilitate that.

I discovered a small bug in this workflow a few years ago and submitted
the following patch to linux-xlnx:
https://github.com/Xilinx/linux-xlnx/pull/106

The first byte of a sub-page read is sometimes corrupted (I.e. READ0
followed by RNDOUT command sequence) because (I suspect) RNDOUT commands
can linger somewhere on the interconnect between the cpu and pl353
during the 100ns delay period, whereas other commands are synchronized
via status change or interrupt to indicate completion. A data store
barrier prior to ndelay() seems to address this issue. It hasn't
reproduced in about 6 months of read-stress testing on a Zynq-7020 SoC
with this change.

Have you considered pulling this change?
If not, I'm curious how (if at all) you'd recommend implementing partial
page reads with the pl353?

Thanks,
Haris

>>
>>> + if (nfc_op.cmnds[0] == 0xef)
>>> + nfc_op.wait = false;
>>> + if (nfc_op.wait) {
>>> + nfc_op.wait = false;
>>
>> You can remove this line.
> Yes, it is initializing to zero in cmd_function.
>>
>>> + do {
>>> + if (chip->dev_ready(mtd))
>>> + break;
>>> + cpu_relax();
>>> + } while (!time_after_eq(jiffies, timeout));
>>> + if (time_after_eq(jiffies, timeout)) {
>>> + pr_err("%s timed out\n", __func__);
>>> + return -ETIMEDOUT;
>>> + }
>>
>> Same comment about the wait.
> Ok, I will update it.
>>
>>> + return 0;
>>> + }
>>> + }
>>> +
>>> + if (instr == NULL)
>>
>> if (!instr)
> I will modify it.
>>
>>> + return 0;
>>> + if (instr->type == NAND_OP_DATA_IN_INSTR)
>>> + return pl353_nand_read_buf(chip, instr->ctx.data.buf.in, len);
>>> +
>>> + if (instr->type == NAND_OP_DATA_OUT_INSTR) {
>>> + if ((nfc_op.cmnds[0] == NAND_CMD_PAGEPROG) ||
>>> + (nfc_op.cmnds[0] == NAND_CMD_SEQIN))
>>> + pl353_nand_write_page_raw(mtd, chip,
>>> + instr->ctx.data.buf.out, 0, nfc_op.addrs);
>>> + else
>>> + pl353_nand_write_buf_l(chip, instr->ctx.data.buf.out,
>>> + len);
>>> + return 0;
>>> + }
>>> + return 0;
>>> +}
>>> +
>>> +static const struct nand_op_parser pl353_nfc_op_parser =
>> NAND_OP_PARSER(
>>> + NAND_OP_PARSER_PATTERN(
>>> + pl353_nand_cmd_function,
>>> + NAND_OP_PARSER_PAT_CMD_ELEM(false),
>>> + NAND_OP_PARSER_PAT_ADDR_ELEM(false, 7),
>>> + NAND_OP_PARSER_PAT_DATA_IN_ELEM(false, 8)),
>>> + NAND_OP_PARSER_PATTERN(
>>> + pl353_nand_cmd_function,
>>> + NAND_OP_PARSER_PAT_CMD_ELEM(false),
>>> + NAND_OP_PARSER_PAT_ADDR_ELEM(false, 7),
>>> + NAND_OP_PARSER_PAT_CMD_ELEM(false),
>>> + NAND_OP_PARSER_PAT_WAITRDY_ELEM(false)),
>>> + NAND_OP_PARSER_PATTERN(
>>> + pl353_nand_cmd_function,
>>> + NAND_OP_PARSER_PAT_CMD_ELEM(false),
>>> + NAND_OP_PARSER_PAT_DATA_IN_ELEM(false, 8)),
>>> + NAND_OP_PARSER_PATTERN(
>>> + pl353_nand_cmd_function,
>>> + NAND_OP_PARSER_PAT_CMD_ELEM(false),
>>> + NAND_OP_PARSER_PAT_ADDR_ELEM(false, 7),
>>> + NAND_OP_PARSER_PAT_WAITRDY_ELEM(false)),
>>> + NAND_OP_PARSER_PATTERN(
>>> + pl353_nand_cmd_function,
>>> + NAND_OP_PARSER_PAT_CMD_ELEM(false),
>>> + NAND_OP_PARSER_PAT_ADDR_ELEM(false, 8),
>>> + NAND_OP_PARSER_PAT_DATA_OUT_ELEM(false, 2048),
>>> + NAND_OP_PARSER_PAT_CMD_ELEM(true),
>>> + NAND_OP_PARSER_PAT_WAITRDY_ELEM(true)),
>>> + NAND_OP_PARSER_PATTERN(
>>> + pl353_nand_cmd_function,
>>> + NAND_OP_PARSER_PAT_CMD_ELEM(false),
>>> + NAND_OP_PARSER_PAT_WAITRDY_ELEM(false)),
>>> + NAND_OP_PARSER_PATTERN(
>>> + pl353_nand_cmd_function,
>>> + NAND_OP_PARSER_PAT_DATA_IN_ELEM(false, 2048)),
>>> + NAND_OP_PARSER_PATTERN(
>>> + pl353_nand_cmd_function,
>>> + NAND_OP_PARSER_PAT_CMD_ELEM(false)),
>>
>> I am pretty sure you can factorize all these patterns now. Use the "optional"
>> parameter for that.
> Can you explain little bit? I didn't get.
>>
>>> + );
>>> +
>>> +static int pl353_nfc_exec_op(struct nand_chip *chip,
>>> + const struct nand_operation *op,
>>> + bool check_only)
>>> +{
>>> + return nand_op_parser_exec_op(chip, &pl353_nfc_op_parser,
>>> + op, check_only);
>>> +}
>>> +
>>> +/**
>>> + * pl353_nand_device_ready - Check device ready/busy line
>>> + * @mtd: Pointer to the mtd_info structure
>>> + *
>>> + * Return: 0 on busy or 1 on ready state
>>> + */
>>> +static int pl353_nand_device_ready(struct mtd_info *mtd) {
>>> + if (pl353_smc_get_nand_int_status_raw()) {
>>
>> I don't understand where this new function is declared.
> Added links above for this driver.
>>
>>> + pl353_smc_clr_nand_int();
>>> + return 1;
>>> + }
>>> + return 0;
>>> +}
>>> +
>>> +/**
>>> + * pl353_nand_ecc_init - Initialize the ecc information as per the ecc mode
>>> + * @mtd: Pointer to the mtd_info structure
>>> + * @ecc: Pointer to ECC control structure
>>> + * @ecc_mode: ondie ecc status
>>> + *
>>> + * This function initializes the ecc block and functional pointers as
>>> +per the
>>> + * ecc mode
>>> + */
>>> +static void pl353_nand_ecc_init(struct mtd_info *mtd, struct nand_ecc_ctrl
>> *ecc,
>>> + int ecc_mode)
>>> +{
>>> + struct nand_chip *chip = mtd_to_nand(mtd);
>>> +
>>> + ecc->read_oob = pl353_nand_read_oob;
>>> + ecc->read_page_raw = pl353_nand_read_page_raw;
>>> + ecc->write_oob = pl353_nand_write_oob;
>>> + ecc->write_page_raw = pl353_nand_write_page_raw;
>>> +
>>> + if (ecc_mode == NAND_ECC_ON_DIE) {
>>> + pl353_smc_set_ecc_mode(PL353_SMC_ECCMODE_BYPASS);
>>> +
>>> + /*
>>> + * The software ECC routines won't work with the
>>> + * SMC controller
>>> + */
>>> + ecc->read_page = pl353_nand_read_page_raw;
>>> + ecc->write_page = pl353_nand_write_page_raw;
>>> + /*
>>> + * On-Die ECC spare bytes offset 8 is used for ECC codes
>>> + * Use the BBT pattern descriptors
>>> + */
>>> + chip->bbt_td = &bbt_main_descr;
>>> + chip->bbt_md = &bbt_mirror_descr;
>>> + } else {
>>> + ecc->mode = NAND_ECC_HW;
>>> + /* Hardware ECC generates 3 bytes ECC code for each 512
>> bytes */
>>> + ecc->bytes = 3;
>>> + ecc->strength = 1;
>>> + ecc->calculate = pl353_nand_calculate_hwecc;
>>> + ecc->correct = pl353_nand_correct_data;
>>> + ecc->hwctl = NULL;
>>
>> You can drop this.
> Ok, I will update it.
>>
>>> + ecc->read_page = pl353_nand_read_page_hwecc;
>>> + ecc->size = PL353_NAND_ECC_SIZE;
>>> + ecc->write_page = pl353_nand_write_page_hwecc;
>>> + pl353_smc_set_ecc_pg_size(mtd->writesize);
>>> + switch (mtd->writesize) {
>>> + case 512:
>>> + case 1024:
>>> + case 2048:
>>
>> I prefer SZ_512, SZ_1K, SZ_2K.
> Ok, I will update it.
> Thanks again.
> If any of the above comments given by you, are not addressed properly,
> Please let me know.
>
> Thanks,
> Naga Sureshkumar Reli.
>>
>>> +
>> pl353_smc_set_ecc_mode(PL353_SMC_ECCMODE_APB);
>>> + break;
>>> + default:
>>> + /*
>>> + * The software ECC routines won't work with the
>>> + * SMC controller
>>> + */
>>> + ecc->calculate = nand_calculate_ecc;
>>> + ecc->correct = nand_correct_data;
>>> + ecc->read_page = pl353_nand_read_page_swecc;
>>> + ecc->write_page = pl353_nand_write_page_swecc;
>>> + ecc->size = 256;
>>> + break;
>>> + }
>>> +
>>> + if (mtd->oobsize == 16)
>>> + mtd_set_ooblayout(mtd,
>> &pl353_ecc_ooblayout16_ops);
>>> + else if (mtd->oobsize == 64)
>>> + mtd_set_ooblayout(mtd,
>> &pl353_ecc_ooblayout64_ops);
>>> + }
>>> +}
>>> +
>>> +/**
>>> + * pl353_nand_probe - Probe method for the NAND driver
>>> + * @pdev: Pointer to the platform_device structure
>>> + *
>>> + * This function initializes the driver data structures and the hardware.
>>> + *
>>> + * Return: 0 on success or error value on failure
>>> + */
>>> +static int pl353_nand_probe(struct platform_device *pdev) {
>>> + struct pl353_nand_info *xnand;
>>> + struct mtd_info *mtd;
>>> + struct nand_chip *nand_chip;
>>> + struct resource *res;
>>> +
>>> + xnand = devm_kzalloc(&pdev->dev, sizeof(*xnand), GFP_KERNEL);
>>> + if (!xnand)
>>> + return -ENOMEM;
>>> +
>>> + /* Map physical address of NAND flash */
>>> + res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
>>> + xnand->nand_base = devm_ioremap_resource(&pdev->dev, res);
>>> + if (IS_ERR(xnand->nand_base))
>>> + return PTR_ERR(xnand->nand_base);
>>> +
>>> + nand_chip = &xnand->chip;
>>> + mtd = nand_to_mtd(nand_chip);
>>> + nand_chip->exec_op = pl353_nfc_exec_op;
>>> + nand_set_controller_data(nand_chip, xnand);
>>> + mtd->priv = nand_chip;
>>> + mtd->owner = THIS_MODULE;
>>> + mtd->name = PL353_NAND_DRIVER_NAME;
>>> + nand_set_flash_node(nand_chip, pdev->dev.of_node);
>>> +
>>> + /* Set address of NAND IO lines */
>>> + nand_chip->IO_ADDR_R = xnand->nand_base;
>>> + nand_chip->IO_ADDR_W = xnand->nand_base;
>>> + /* Set the driver entry points for MTD */
>>> + nand_chip->dev_ready = pl353_nand_device_ready;
>>> + nand_chip->select_chip = pl353_nand_select_chip;
>>> + /* If we don't set this delay driver sets 20us by default */
>>> + nand_chip->chip_delay = 30;
>>> +
>>> + /* Set the device option and flash width */
>>> + nand_chip->options = NAND_BUSWIDTH_AUTO;
>>> + nand_chip->bbt_options = NAND_BBT_USE_FLASH;
>>> + platform_set_drvdata(pdev, xnand);
>>> +
>>> + /* first scan to find the device and get the page size */
>>> + if (nand_scan_ident(mtd, 1, NULL)) {
>>> + dev_err(&pdev->dev, "nand_scan_ident for NAND failed\n");
>>> + return -ENXIO;
>>> + }
>>> +
>>> + xnand->row_addr_cycles = nand_chip->onfi_params.addr_cycles & 0xF;
>>> + xnand->col_addr_cycles =
>>> + (nand_chip->onfi_params.addr_cycles >> 4) &
>> 0xF;
>>> +
>>> + pl353_nand_ecc_init(mtd, &nand_chip->ecc, nand_chip->ecc.mode);
>>> + if (nand_chip->options & NAND_BUSWIDTH_16)
>>> + pl353_smc_set_buswidth(PL353_SMC_MEM_WIDTH_16);
>>> + /* second phase scan */
>>> + if (nand_scan_tail(mtd)) {
>>> + dev_err(&pdev->dev, "nand_scan_tail for NAND failed\n");
>>> + return -ENXIO;
>>> + }
>>> +
>>> + mtd_device_register(mtd, NULL, 0);
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +/**
>>> + * pl353_nand_remove - Remove method for the NAND driver
>>> + * @pdev: Pointer to the platform_device structure
>>> + *
>>> + * This function is called if the driver module is being unloaded. It
>>> +frees all
>>> + * resources allocated to the device.
>>> + *
>>> + * Return: 0 on success or error value on failure
>>> + */
>>> +static int pl353_nand_remove(struct platform_device *pdev) {
>>> + struct pl353_nand_info *xnand = platform_get_drvdata(pdev);
>>> + struct mtd_info *mtd = nand_to_mtd(&xnand->chip);
>>> +
>>> + /* Release resources, unregister device */
>>> + nand_release(mtd);
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +/* Match table for device tree binding */ static const struct
>>> +of_device_id pl353_nand_of_match[] = {
>>> + { .compatible = "arm,pl353-nand-r2p1" },
>>> + {},
>>> +};
>>> +MODULE_DEVICE_TABLE(of, pl353_nand_of_match);
>>> +
>>> +/*
>>> + * pl353_nand_driver - This structure defines the NAND subsystem
>>> +platform driver */ static struct platform_driver pl353_nand_driver =
>>> +{
>>> + .probe = pl353_nand_probe,
>>> + .remove = pl353_nand_remove,
>>> + .driver = {
>>> + .name = PL353_NAND_DRIVER_NAME,
>>> + .of_match_table = pl353_nand_of_match,
>>> + },
>>> +};
>>> +
>>> +module_platform_driver(pl353_nand_driver);
>>> +
>>> +MODULE_AUTHOR("Xilinx, Inc.");
>>> +MODULE_ALIAS("platform:" PL353_NAND_DRIVER_NAME);
>>> +MODULE_DESCRIPTION("ARM PL353 NAND Flash Driver");
>>> +MODULE_LICENSE("GPL");
>>
>> Thanks,
>> Miquèl
>>
>> --
>> Miquel Raynal, Bootlin (formerly Free Electrons) Embedded Linux and Kernel
>> engineering https://urldefense.proofpoint.com/v2/url?u=https-3A__bootlin.com&d=DwIGaQ&c=I_0YwoKy7z5LMTVdyO6YCiE2uzI1jjZZuIPelcSjixA&r=8Bziuw3IaCGjyrSAphuGwHmVdHcVwza-srUYwL9U_Ms&m=Qud5yLYSsZO8CluBVnAoEJUNWNJcqCra13xxjmbxyxs&s=dSALo-GHPaytbxPIGQKjdbPVsHzRNSve8iZatjQlIF0&e=

\
 
 \ /
  Last update: 2018-04-06 21:23    [W:0.248 / U:0.232 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site