mtd: nand: omap: optimize OMAP NAND driver
Subsystems (3)
v9 cover mtd: nand: omap: optimize OMAP NAND driver
OMAP NAND drivers support various ecc-schemes like (HAM1, BCH4, BCH8..) And, each ecc-scheme usually has 2 implementations
- HAM1_HW or BCHx_HW (using in-build h/w engine for all computation)
- HAM1_SW or BCHx_HW_DETECTION_SW) (using software library for some computation)
This patch series is Part-2 (split) from earlier series, as mentioned below:
http://lists.denx.de/pipermail/u-boot/2013-November/167270.html
This series aims at following:
- cleans-up redundant code, duplicate in different ecc-schemes
- optimizes ecc calculation and correction paths
- improves scalability and readability, for adding newer ecc-schemes in future.
- updates following callbacks for different ecc-schmes:
[PATCH 1/4] nand_chip->ecc.hwctl(): configures h/w controller before for Read/Write acccess [PATCH 2/4] nand_chip->ecc.calculate(): calculates and fetches ECC syndrome from h/w controller. [PATCH 3/4] nand_chip->ecc.correct(): detects and corrects ECC errors on read access [PATCH 4/4] update doc and README
*changes in v9*
- split from main series, rebased and forked independently
*changes in v8*
http://lists.denx.de/pipermail/u-boot/2013-October/164648.html
- minor code clean-up
*changes in v7*
[PATCH 1/4] http://lists.denx.de/pipermail/u-boot/2013-September/163868.html
- removed: re-configuration of gpmc.config1[dev_width] added in previous version of patch
*changes in v6*
http://lists.denx.de/pipermail/u-boot/2013-September/163099.html
- cosmetic changes
*changes in v5*
[PATCH 1/4] http://lists.denx.de/pipermail/u-boot/2013-September/162105.html
- fixed device-width in GPMC_CONFIG1_X to support x16 devices
[PATCH 2/4] http://lists.denx.de/pipermail/u-boot/2013-September/162109.html
- code clean-up for OMAP_ECC_BCH8_CODE_HW_DETECTION_SW mode
[PATCH 3/4] http://lists.denx.de/pipermail/u-boot/2013-September/162110.html
- fixed omap_correct_data_bch() for correcting bit-flips using ELM
- code-cleanup + added omap_reverse_list()
[PATCH 4/4] incorporated feedbacks from Peter Korsgaard <[email protected]>
*changes in v4*
http://lists.denx.de/pipermail/u-boot/2013-September/161861.html
- minor clean-up
*changes in v3*
http://lists.denx.de/pipermail/u-boot/2013-August/161665.html
- removed un-used defines from asm/omap_gpmc.h
- removed omap_calculate_ecc_bch_sw() and omap_calculate_ecc() and merged their logic into omap_calculate_ecc_bch()
- updated return error-code in-case of invalid conditions
- (new) for board/ti/am335x/README
*changes in v2*
http://lists.denx.de/pipermail/u-boot/2013-August/160749.html
- fixed board_nand_init() and omap_enable_hwecc()
*original v1*
http://lists.denx.de/pipermail/u-boot/2013-August/160218.html
Pekon Gupta (4):
mtd: nand: omap: optimize chip->ecc.hwctl() for H/W ECC schemes
mtd: nand: omap: optimize chip->ecc.calculate() for H/W ECC schemes
mtd: nand: omap: optimized chip->ecc.correct() for H/W ECC schemes board/ti/am335x/README: update for NAND boot
arch/arm/include/asm/omap_gpmc.h | 7 - board/ti/am335x/README | 53 ++-- drivers/mtd/nand/omap_gpmc.c | 565 +++++++++++++-------------------------- 3 files changed, 217 insertions(+), 408 deletions(-)
Signature · 1 line
1.8.1
PATCH v9 1/4 mtd: nand: omap: optimize chip->ecc.hwctl() for H/W ECC schemes
chip->ecc.hwctl() is used for preparing the H/W controller before read/write NAND accesses (like assigning data-buf, enabling ECC scheme configs, etc.)
Though all ECC schemes in OMAP NAND driver use GPMC controller for generating ECC syndrome (for both Read/Write accesses). But but in current code HAM1_ECC and BCHx_ECC schemes implement individual function to achieve this. This patch
(1) removes omap_hwecc_init() and omap_hwecc_init_bch()
as chip->ecc.hwctl will re-initializeGPMC before every read/write call. omap_hwecc_init_bch() -> omap_enable_ecc_bch()
(2) merges the GPMC configuration code for all ECC schemes into single omap_enable_hwecc(), thus adding scalability for future ECC schemes. omap_enable_hwecc() + omap_enable_ecc_bch() -> omap_enable_hwecc()
--- arch/arm/include/asm/omap_gpmc.h | 7 -- drivers/mtd/nand/omap_gpmc.c | 192 +++++++++++---------------------------- 2 files changed, 54 insertions(+), 145 deletions(-)
arch/arm/include/asm/omap_gpmc.h +0 -7
@@ -14,13 +14,6 @@#define GPMC_BUF_EMPTY 0#define GPMC_BUF_FULL 1 -#define ECCCLEAR (0x1 << 8) -#define ECCRESULTREG1 (0x1 << 0) -#define ECCSIZE512BYTE 0xFF -#define ECCSIZE1 (ECCSIZE512BYTE << 22) -#define ECCSIZE0 (ECCSIZE512BYTE << 12) -#define ECCSIZE0SEL (0x000 << 0) -/* Generic ECC Layouts */ /* Large Page x8 NAND device Layout */ #ifdef GPMC_NAND_ECC_LP_x8_LAYOUT
drivers/mtd/nand/omap_gpmc.c +54 -138
243 lines not on this page · raw message
Signature · 1 line
1.8.1
PATCH v9 2/4 mtd: nand: omap: optimize chip->ecc.calculate() for H/W ECC schemes
In reply to Pekon Guptachip->ecc.calculate() is used for calculating and fetching of ECC syndrome by processing the data passed during Read/Write accesses.
All H/W based ECC schemes use GPMC controller to calculate ECC syndrome. But each BCHx_ECC scheme has its own implemetation of post-processing and fetching ECC syndrome from GPMC controller.
This patch updates OMAP_ECC_BCH8_CODE_HW ECC scheme in following way:
- merges multiple chip->calculate API for different ECC schemes omap_calculate_ecc() + omap_calculate_ecc_bch() + omap_calculate_ecc_bch_sw()
==> omap_calculate_ecc()
- removes omap_ecc_disable() and instead uses it as inline.
--- drivers/mtd/nand/omap_gpmc.c | 226 ++++++++++++------------------------------- 1 file changed, 63 insertions(+), 163 deletions(-)
drivers/mtd/nand/omap_gpmc.c +63 -163
@@ -21,7 +21,10 @@#define SECTOR_BYTES 512#define ECCCLEAR (0x1 << 8)#define ECCRESULTREG1 (0x1 << 0) - +#ifdef CONFIG_BCH +static u8 bch8_polynomial[] = {0xef ,0x51 ,0x2e ,0x09 ,0xed ,0x93 ,0x9a ,0xc2 , +0x97 ,0x79 ,0xe5 ,0x24 ,0xb5 }; +#endif static uint8_t cs;static __maybe_unused struct nand_ecclayout omap_ecclayout;
@@ -143,42 +146,6 @@ static int __maybe_unused omap_correct_data(struct mtd_info *mtd, uint8_t *dat, }/* -* omap_calculate_ecc - Generate non-inverted ECC bytes. -* -* Using noninverted ECC can be considered ugly since writing a blank -* page ie. padding will clear the ECC bytes. This is no problem as -* long nobody is trying to write data on the seemingly unused page. -* Reading an erased page will produce an ECC mismatch between -* generated and read ECC bytes that has to be dealt with separately. -* E.g. if page is 0xFF (fresh erased), and if HW ECC engine within GPMC -* is used, the result of read will be 0x0 while the ECC offsets of the -* spare area will be 0xFF which will result in an ECC mismatch. -* @mtd: MTD structure -* @dat: unused -* @ecc_code: ecc_code buffer -*/ -static int __maybe_unused omap_calculate_ecc (struct mtd_info *mtd, -const uint8_t *dat,uint8_t *ecc_code) -{ -u_int32_t val; - -/* Start Reading from HW ECC1_Result = 0x200 */ - val =readl (&gpmc_cfg->ecc1_result); - - ecc_code[0 ] = val &0xFF ; - ecc_code[1 ] = (val >>16 ) &0xFF ; - ecc_code[2 ] = ((val >>8 ) &0x0F ) | ((val >>20 ) &0xF0 ); - -/* -* Stop reading anymore ECC vals and clear old results -* enable will be called if more reads are required -*/ -writel (0x000 , &gpmc_cfg->ecc_config); - -return 0 ; -} - -/* * Generic BCH interface */ struct nand_bch_priv {
@@ -194,12 +161,7 @@ struct nand_bch_priv {#define ECC_BCH8 1#define ECC_BCH16 2 -/* GPMC ecc engine settings */ -#define BCH_WRAPMODE_1 1/* BCH wrap mode 1 */ -#define BCH_WRAPMODE_6 6/* BCH wrap mode 6 */ -/* BCH nibbles for diff bch levels */ -#define NAND_ECC_HW_BCH ((uint8_t)(NAND_ECC_HW_OOB_FIRST) + 1)#define ECC_BCH4_NIBBLES 13#define ECC_BCH8_NIBBLES 26#define ECC_BCH16_NIBBLES 52
@@ -211,7 +173,6 @@ struct nand_bch_priv {* When some users with other BCH strength will exists this have to change! */ static __maybe_unused struct nand_bch_priv bch_priv = { - .mode =NAND_ECC_HW_BCH , .type =ECC_BCH8 , .nibbles =ECC_BCH8_NIBBLES , .control =NULL
@@ -280,57 +241,76 @@ static void omap_enable_hwecc(struct mtd_info *mtd, int32_t mode) }/* -* omap_ecc_disable - Disable H/W ECC calculation -* -* @mtd: MTD device structure -*/ -static void __maybe_unused omap_ecc_disable (struct mtd_info *mtd) -{ -writel ((readl (&gpmc_cfg->ecc_config) & ~0x1 ), &gpmc_cfg->ecc_config); -} - -/* -* BCH support using ELM module -*/ -#ifdef CONFIG_NAND_OMAP_ELM -/* -* omap_read_bch8_result - Read BCH result for BCH8 level -* -* @mtd: MTD device structure -* @big_endian: When set read register 3 first -* @ecc_code: Read syndrome from BCH result registers +* omap_calculate_ecc - Read ECC result +* @mtd: MTD structure +* @dat: unused +* @ecc_code: ecc_code buffer +* Using noninverted ECC can be considered ugly since writing a blank +* page ie. padding will clear the ECC bytes. This is no problem as +* long nobody is trying to write data on the seemingly unused page. +* Reading an erased page will produce an ECC mismatch between +* generated and read ECC bytes that has to be dealt with separately. +* E.g. if page is 0xFF (fresh erased), and if HW ECC engine within GPMC +* is used, the result of read will be 0x0 while the ECC offsets of the +* spare area will be 0xFF which will result in an ECC mismatch. */ -static void omap_read_bch8_result (struct mtd_info *mtd,uint8_t big_endian, +static int omap_calculate_ecc (struct mtd_info *mtd,const uint8_t *dat,uint8_t *ecc_code) { -uint32_t *ptr; +struct nand_chip *chip = mtd->priv; +struct nand_bch_priv *bch = chip->priv; +uint32_t *ptr, val =0 ;int8_t i =0 , j; -if (big_endian) { +switch (bch->ecc_scheme) { +case OMAP_ECC_HAM1_CODE_HW : + val =readl (&gpmc_cfg->ecc1_result); + ecc_code[0 ] = val &0xFF ; + ecc_code[1 ] = (val >>16 ) &0xFF ; + ecc_code[2 ] = ((val >>8 ) &0x0F ) | ((val >>20 ) &0xF0 ); +break ; +#ifdef CONFIG_BCH +case OMAP_ECC_BCH8_CODE_HW_DETECTION_SW : +#endif +case OMAP_ECC_BCH8_CODE_HW : ptr = &gpmc_cfg->bch_result_0_3[0 ].bch_result_x[3 ]; - ecc_code[i++] =readl (ptr) &0xFF ; + val =readl (ptr); + ecc_code[i++] = (val >>0 ) &0xFF ; ptr--;for (j =0 ; j <3 ; j++) { - ecc_code[i++] = (readl (ptr) >>24 ) &0xFF ; - ecc_code[i++] = (readl (ptr) >>16 ) &0xFF ; - ecc_code[i++] = (readl (ptr) >>8 ) &0xFF ; - ecc_code[i++] =readl (ptr) &0xFF ; + val =readl (ptr); + ecc_code[i++] = (val >>24 ) &0xFF ; + ecc_code[i++] = (val >>16 ) &0xFF ; + ecc_code[i++] = (val >>8 ) &0xFF ; + ecc_code[i++] = (val >>0 ) &0xFF ; ptr--; } - }else { - ptr = &gpmc_cfg->bch_result_0_3[0 ].bch_result_x[0 ]; -for (j =0 ; j <3 ; j++) { - ecc_code[i++] =readl (ptr) &0xFF ; - ecc_code[i++] = (readl (ptr) >>8 ) &0xFF ; - ecc_code[i++] = (readl (ptr) >>16 ) &0xFF ; - ecc_code[i++] = (readl (ptr) >>24 ) &0xFF ; - ptr++; - } - ecc_code[i++] =readl (ptr) &0xFF ; - ecc_code[i++] =0 ;/* 14th byte is always zero */ +break ; +default : +return -EINVAL ; + } +/* ECC scheme specific syndrome customizations */ +switch (bch->ecc_scheme) { +case OMAP_ECC_HAM1_CODE_HW : +break ; +#ifdef CONFIG_BCH +case OMAP_ECC_BCH8_CODE_HW_DETECTION_SW : + +for (i =0 ; i < chip->ecc.bytes; i++) + *(ecc_code + i) = *(ecc_code + i) ^ + bch8_polynomial[i]; +break ; +#endif +case OMAP_ECC_BCH8_CODE_HW : + ecc_code[chip->ecc.bytes -1 ] =0x00 ; +break ; +default : +return -EINVAL ; } +return 0 ; } +#ifdef CONFIG_NAND_OMAP_ELM /* * omap_rotate_ecc_bch - Rotate the syndrome bytes *
@@ -367,35 +347,6 @@ static void omap_rotate_ecc_bch(struct mtd_info *mtd, uint8_t *calc_ecc, }/* -* omap_calculate_ecc_bch - Read BCH ECC result -* -* @mtd: MTD structure -* @dat: unused -* @ecc_code: ecc_code buffer -*/ -static int omap_calculate_ecc_bch (struct mtd_info *mtd,const uint8_t *dat, -uint8_t *ecc_code) -{ -struct nand_chip *chip = mtd->priv; -struct nand_bch_priv *bch = chip->priv; -uint8_t big_endian =1 ; -int8_t ret =0 ; - -if (bch->type ==ECC_BCH8 ) -omap_read_bch8_result (mtd, big_endian, ecc_code); -else /* BCH4 and BCH16 currently not supported */ - ret =-1 ; - -/* -* Stop reading anymore ECC vals and clear old results -* enable will be called if more reads are required -*/ -omap_ecc_disable (mtd); - -return ret; -} - -/* * omap_fix_errors_bch - Correct bch error in the data * * @mtd: MTD device structure
@@ -551,57 +502,6 @@ static int omap_read_page_bch(struct mtd_info *mtd, struct nand_chip *chip,* OMAP3 BCH8 support (with BCH library) */ #ifdef CONFIG_BCH -/* -* omap_calculate_ecc_bch_sw - Read BCH ECC result -* -* @mtd: MTD device structure -* @dat: The pointer to data on which ecc is computed (unused here) -* @ecc: The ECC output buffer -*/ -static int omap_calculate_ecc_bch_sw (struct mtd_info *mtd,const uint8_t *dat, -uint8_t *ecc) -{ -int ret =0 ; -size_t i; -unsigned long nsectors, val1, val2, val3, val4; - - nsectors = ((readl (&gpmc_cfg->ecc_config) >>4 ) &0x7 ) +1 ; - -for (i =0 ; i < nsectors; i++) { -/* Read hw-computed remainder */ - val1 =readl (&gpmc_cfg->bch_result_0_3[i].bch_result_x[0 ]); - val2 =readl (&gpmc_cfg->bch_result_0_3[i].bch_result_x[1 ]); - val3 =readl (&gpmc_cfg->bch_result_0_3[i].bch_result_x[2 ]); - val4 =readl (&gpmc_cfg->bch_result_0_3[i].bch_result_x[3 ]); - -/* -* Add constant polynomial to remainder, in order to get an ecc -* sequence of 0xFFs for a buffer filled with 0xFFs. -*/ - *ecc++ =0xef ^ (val4 &0xFF ); - *ecc++ =0x51 ^ ((val3 >>24 ) &0xFF ); - *ecc++ =0x2e ^ ((val3 >>16 ) &0xFF ); - *ecc++ =0x09 ^ ((val3 >>8 ) &0xFF ); - *ecc++ =0xed ^ (val3 &0xFF ); - *ecc++ =0x93 ^ ((val2 >>24 ) &0xFF ); - *ecc++ =0x9a ^ ((val2 >>16 ) &0xFF ); - *ecc++ =0xc2 ^ ((val2 >>8 ) &0xFF ); - *ecc++ =0x97 ^ (val2 &0xFF ); - *ecc++ =0x79 ^ ((val1 >>24 ) &0xFF ); - *ecc++ =0xe5 ^ ((val1 >>16 ) &0xFF ); - *ecc++ =0x24 ^ ((val1 >>8 ) &0xFF ); - *ecc++ =0xb5 ^ (val1 &0xFF ); - } - -/* -* Stop reading anymore ECC vals and clear old results -* enable will be called if more reads are required -*/ -omap_ecc_disable (mtd); - -return ret; -} -/** * omap_correct_data_bch_sw - Decode received data and correct errors * @mtd: MTD device structure
@@ -746,7 +646,7 @@ static int omap_select_ecc_scheme(struct nand_chip *nand, nand->ecc.bytes =13 ; nand->ecc.hwctl = omap_enable_hwecc; nand->ecc.correct = omap_correct_data_bch_sw; - nand->ecc.calculate = omap_calculate_ecc_bch_sw; + nand->ecc.calculate = omap_calculate_ecc;/* define ecc-layout */ ecclayout->eccbytes = nand->ecc.bytes * eccsteps; ecclayout->eccpos[0 ] =BADBLOCK_MARKER_LENGTH ;
@@ -787,7 +687,7 @@ static int omap_select_ecc_scheme(struct nand_chip *nand, nand->ecc.bytes =14 ; nand->ecc.hwctl = omap_enable_hwecc; nand->ecc.correct = omap_correct_data_bch; - nand->ecc.calculate = omap_calculate_ecc_bch; + nand->ecc.calculate = omap_calculate_ecc; nand->ecc.read_page = omap_read_page_bch;/* define ecc-layout */ ecclayout->eccbytes = nand->ecc.bytes * eccsteps;
Signature · 1 line
1.8.1
PATCH v9 3/4 mtd: nand: omap: optimized chip->ecc.correct() for H/W ECC schemes
In reply to Pekon Guptachip->ecc.correct() is used for detecting and correcting bit-flips during read operations. In omap-nand driver it implemented as:
(a) omap_correct_data(): for h/w based ECC_HAM1 scheme
(b) omap_correct_data_bch() + CONFIG_NAND_OMAP_ECC_BCH8_CODE_HW_DETECTION_SW for ECC_BCH8 scheme using GPMC and software lib/bch.c
(c) omap_correct_data_bch() + CONFIG_NAND_OMAP_ECC_BCH8_CODE_HW for ECC_BCH8 scheme using GPMC and ELM
This patch updates (c)
- checks for calc_ecc[]==0x00 so that error_correction is not required for known good pages.
- adds scalability for other ECC_BCHx scheme by merging following omap_rotate_ecc_bch() + omap_fix_errors_bch() => omap_correct_data_bch()
- fixing logic for bit-flip correction based on error_loc[count]
--- drivers/mtd/nand/omap_gpmc.c | 161 +++++++++++++++++++------------------------ 1 file changed, 71 insertions(+), 90 deletions(-)
drivers/mtd/nand/omap_gpmc.c +71 -90
@@ -21,6 +21,9 @@#define SECTOR_BYTES 512#define ECCCLEAR (0x1 << 8)#define ECCRESULTREG1 (0x1 << 0) +/* 4 bit padding to make byte aligned, 56 = 52 + 4 */ +#define BCH4_BIT_PAD 4 +#ifdef CONFIG_BCH static u8 bch8_polynomial[] = {0xef ,0x51 ,0x2e ,0x09 ,0xed ,0x93 ,0x9a ,0xc2 ,0x97 ,0x79 ,0xe5 ,0x24 ,0xb5 };
@@ -179,6 +182,23 @@ static __maybe_unused struct nand_bch_priv bch_priv = { };/* +* omap_reverse_list - re-orders list elements in reverse order [internal] +* @list: pointer to start of list +* @length: length of list +*/ +void omap_reverse_list (u8 *list,unsigned int length) +{ +unsigned int i, j; +unsigned int half_length = length /2 ; +u8 tmp; +for (i =0 , j = length -1 ; i < half_length; i++, j--) { + tmp = list[i]; + list[i] = list[j]; + list[j] = tmp; + } +} + +/* * omap_enable_hwecc - configures GPMC as per ECC scheme before read/write * @mtd: MTD device structure * @mode: Read/Write mode
@@ -312,77 +332,6 @@ static int omap_calculate_ecc(struct mtd_info *mtd, const uint8_t *dat,#ifdef CONFIG_NAND_OMAP_ELM /* -* omap_rotate_ecc_bch - Rotate the syndrome bytes -* -* @mtd: MTD device structure -* @calc_ecc: ECC read from ECC registers -* @syndrome: Rotated syndrome will be retuned in this array -* -*/ -static void omap_rotate_ecc_bch (struct mtd_info *mtd,uint8_t *calc_ecc, -uint8_t *syndrome) -{ -struct nand_chip *chip = mtd->priv; -struct nand_bch_priv *bch = chip->priv; -uint8_t n_bytes =0 ; -int8_t i, j; - -switch (bch->type) { -case ECC_BCH4 : - n_bytes =8 ; -break ; - -case ECC_BCH16 : - n_bytes =28 ; -break ; - -case ECC_BCH8 : -default : - n_bytes =13 ; -break ; - } - -for (i =0 , j = (n_bytes-1 ); i < n_bytes; i++, j--) - syndrome[i] = calc_ecc[j]; -} - -/* -* omap_fix_errors_bch - Correct bch error in the data -* -* @mtd: MTD device structure -* @data: Data read from flash -* @error_count:Number of errors in data -* @error_loc: Locations of errors in the data -* -*/ -static void omap_fix_errors_bch (struct mtd_info *mtd,uint8_t *data, -uint32_t error_count,uint32_t *error_loc) -{ -struct nand_chip *chip = mtd->priv; -struct nand_bch_priv *bch = chip->priv; -uint8_t count =0 ; -uint32_t error_byte_pos; -uint32_t error_bit_mask; -uint32_t last_bit = (bch->nibbles *4 ) -1 ; - -/* Flip all bits as specified by the error location array. */ -/* FOR( each found error location flip the bit ) */ -for (count =0 ; count < error_count; count++) { -if (error_loc[count] > last_bit) { -/* Remove the ECC spare bits from correction. */ - error_loc[count] -= (last_bit +1 ); -/* Offset bit in data region */ - error_byte_pos = ((512 *8 ) - - (error_loc[count]) -1 ) /8 ; -/* Error Bit mask */ - error_bit_mask =0x1 << (error_loc[count] %8 ); -/* Toggle the error bit to make the correction. */ - data[error_byte_pos] ^= error_bit_mask; - } - } -} - -/* * omap_correct_data_bch - Compares the ecc read from nand spare area * with ECC registers values and corrects one bit error if it has occured *
@@ -398,40 +347,72 @@ static int omap_correct_data_bch(struct mtd_info *mtd, uint8_t *dat, {struct nand_chip *chip = mtd->priv;struct nand_bch_priv *bch = chip->priv; -uint8_t syndrome[28 ]; -uint32_t error_count =0 ; +uint32_t eccbytes = chip->ecc.bytes; +uint32_t error_count =0 , error_max;uint32_t error_loc[8 ]; -uint32_t i, ecc_flag; +uint32_t i, ecc_flag =0 ; +uint8_t count, err =0 ; +uint32_t byte_pos, bit_pos; +/* check calculated ecc */ +for (i =0 ; i < chip->ecc.bytes && !ecc_flag; i++) { +if (calc_ecc[i] !=0x00 ) + ecc_flag =1 ; + } +if (!ecc_flag) +return 0 ; + +/* check for whether its a erased-page */ ecc_flag =0 ; -for (i =0 ; i < chip->ecc.bytes; i++) +for (i =0 ; i < chip->ecc.bytes && !ecc_flag; i++) {if (read_ecc[i] !=0xff ) ecc_flag =1 ; - + }if (!ecc_flag)return 0 ; -elm_reset (); -elm_config ((enum bch_level )(bch->type)); -/* * while reading ECC result we read it in big endian. * Hence while loading to ELM we have rotate to get the right endian. */ -omap_rotate_ecc_bch (mtd, calc_ecc, syndrome); - +switch (bch->ecc_scheme) { +case OMAP_ECC_BCH8_CODE_HW : +omap_reverse_list (calc_ecc, eccbytes -1 ); +break ; +default : +return -EINVAL ; + }/* use elm module to check for errors */ -if (elm_check_error (syndrome, bch->nibbles, &error_count, - error_loc) !=0 ) { -printf ("ECC: uncorrectable.\n" ); -return -1 ; +elm_config ((enum bch_level )(bch->type)); +if (elm_check_error (calc_ecc, bch->nibbles, &error_count, error_loc)) { +printf ("nand: error: uncorrectable ECC errors\n" ); +return -EINVAL ; } -/* correct bch error */ -if (error_count >0 ) -omap_fix_errors_bch (mtd, dat, error_count, error_loc); - -return 0 ; +for (count =0 ; count < error_count; count++) { +switch (bch->type) { +case ECC_BCH8 : +/* 14th byte in ECC is reserved to match ROM layout */ + error_max =SECTOR_BYTES + (eccbytes -1 ); +break ; +default : +return -EINVAL ; + } + byte_pos = error_max - (error_loc[count] /8 ) -1 ; + bit_pos = error_loc[count] %8 ; +if (byte_pos <SECTOR_BYTES ) { + dat[byte_pos] ^=1 << bit_pos; +printf ("nand: bit-flip corrected @data=%d\n" , byte_pos); + }else if (byte_pos < error_max) { + read_ecc[byte_pos -SECTOR_BYTES ] =1 << bit_pos; +printf ("nand: bit-flip corrected @oob=%d\n" , byte_pos - +SECTOR_BYTES ); + }else { + err = -EBADMSG ; +printf ("nand: error: invalid bit-flip location\n" ); + } + } +return (err) ? err : error_count; }/**
Signature · 1 line
1.8.1
PATCH v9 4/4 board/ti/am335x/README: update for NAND boot
In reply to Pekon GuptaNAND boot mode on AM335x EVM has been verified, and steps to use it has been documented and update in this README
--- board/ti/am335x/README | 53 ++++++++++++++++++++++++++++++++++---------------- 1 file changed, 36 insertions(+), 17 deletions(-)
board/ti/am335x/README +36 -17
@@ -46,23 +46,42 @@ NAND The AM335x GP EVM ships with a 256MiB NAND available in most profiles. In this example to program the NAND we assume that an SD card has been inserted with the files to write in the first SD slot and that mtdparts -have been configured correctly for the board. As a time saving measure we -load MLO into memory in one location, copy it into the three locatations -that the ROM checks for additional valid copies, then load U-Boot into -memory. We then write that whole section of memory to NAND. - -U-Boot # mmc rescan -U-Boot # env default -f -a -U-Boot # nand erase.chip -U-Boot # saveenv -U-Boot # load mmc 0 81000000 MLO -U-Boot # cp.b 81000000 81020000 20000 -U-Boot # cp.b 81000000 81040000 20000 -U-Boot # cp.b 81000000 81060000 20000 -U-Boot # load mmc 0 81080000 u-boot.img -U-Boot # nand write 81000000 0 260000 -U-Boot # load mmc 0 ${loadaddr} uImage -U-Boot # nand write ${loadaddr} kernel 500000 +have been configured correctly for the board. All images are first loaded +into memory, then written to NAND. + +Step-1: Building u-boot for NAND boot + Set following CONFIGxx options for NAND device. + CONFIG_SYS_NAND_PAGE_SIZE number of main bytes in NAND page + CONFIG_SYS_NAND_OOBSIZE number of OOB bytes in NAND page + CONFIG_SYS_NAND_BLOCK_SIZE number of bytes in NAND erase-block + CONFIG_SYS_NAND_ECCPOS ECC map for NAND page + CONFIG_NAND_OMAP_ECCSCHEME (refer doc/README.nand) + +Step-2: Flashing NAND via MMC/SD + # select BOOTSEL to MMC/SD boot and boot from MMC/SD card + U-Boot # mmc rescan + # erase flash + U-Boot # nand erase.chip + U-Boot # env default -f -a + U-Boot # saveenv + # flash MLO. Redundant copies of MLO are kept for failsafe + U-Boot # load mmc 0 0x82000000 MLO + U-Boot # nand write 0x82000000 0x00000 0x20000 + U-Boot # nand write 0x82000000 0x20000 0x20000 + U-Boot # nand write 0x82000000 0x40000 0x20000 + U-Boot # nand write 0x82000000 0x60000 0x20000 + # flash u-boot.img + U-Boot # load mmc 0 0x82000000 u-boot.img + U-Boot # nand write 0x82000000 0x80000 0x60000 + # flash kernel image + U-Boot # load mmc 0 0x82000000 uImage + U-Boot # nand write 0x82000000 ${nandsrcaddr} ${nandimgsize} + # flash filesystem image + U-Boot # load mmc 0 0x82000000 filesystem.img + U-Boot # nand write 0x82000000 ${loadaddress} 0x300000 + +Step-3: Set BOOTSEL pin to select NAND boot, and POR the device. + The device should boot from images flashed on NAND device. NOR ===
Signature · 1 line
1.8.1
v9 cover mtd: nand: omap: optimize OMAP NAND driver
In reply to Pekon GuptaQuoted 7 lines · Pekon Gupta
OMAP NAND drivers support various ecc-schemes like (HAM1, BCH4, BCH8..) And, each ecc-scheme usually has 2 implementations
-HAM1_HWorBCHx_HW(using in-build h/w engine for all computation)
-HAM1_SWorBCHx_HW_DETECTION_SW) (using software library for some
computation)
If I make similar changes as these patches suggest, would I be able to enable NAND support for OMAP4-based boards as well? (OMAP 4460) Or are some additional changes needed to support NAND on the next OMAP version?
-Abraham V.
Gupta
v9 cover mtd: nand: omap: optimize OMAP NAND driver
Hi,
Quoted 7 lines
From: Abraham V. [abraham.varricatt at vvdntech.com]
Quoted 5 lines
Pekon Gupta <[email protected]> wrote:
OMAP NAND drivers support various ecc-schemes like (HAM1, BCH4, BCH8..) And, each ecc-scheme usually has 2 implementations
-HAM1_HWorBCHx_HW(using in-build h/w engine for all computation)
-HAM1_SWorBCHx_HW_DETECTION_SW) (using software library for some computation)
Quoted 5 lines
If I make similar changes as these patches suggest, would I be able to enable NAND support for OMAP4-based boards as well? (OMAP 4460) Or are some additional changes needed to support NAND on the next OMAP version?
drivers/mtd/nand/omap-gpmc.c is common driver for all OMAP and AMxxxx platforms like AM33xx, OMAP3, OMAP4, OMAP5.
So these patches would work with OMAP4 also.
You need to just enable GPMC and ELM drivers for OMAP4 board. And update your board file.
- #define CONFIG_NAND_OMAP_GPMC: enables GPMC driver
- #define CONFIG_NAND_OMAP_ELM: enables ELM driver
As there is no OMAP4 board with NAND flash on it, so these driver are not enabled by default.
Following steps should help you proceed...
Step-1: As reference copy whatever is under 'CONFIG_NAND' in include/configs/am335x_evm.h include/configs/<your_board.h>
Step-2: update CONFIG_xx_NAND_yy for device on your board
Step-3: add pin-mux for NAND.
Step-4: update arch/arm/cpu/armv7/omap-common/mem-common.c
(if required)
with regards, pekon
v9 cover mtd: nand: omap: optimize OMAP NAND driver
In reply to Pekon GuptaQuoted 6 lines · Pekon Gupta
OMAP NAND drivers support various ecc-schemes like (HAM1, BCH4, BCH8..) And, each ecc-scheme usually has 2 implementations
-HAM1_HWorBCHx_HW(using in-build h/w engine for all computation)
-HAM1_SWorBCHx_HW_DETECTION_SW) (using software library for some computation)
[..]
Quoted 8 lines
Pekon Gupta (4):
mtd: nand: omap: optimizechip->ecc.hwctl()for H/W ECC schemes
mtd: nand: omap: optimizechip->ecc.calculate()for H/W ECC schemes
mtd: nand: omap: optimizedchip->ecc.correct()for H/W ECC schemes board/ti/am335x/README: update for NAND boot
Quoted 5 lines
arch/arm/include/asm/omap_gpmc.h | 7 - board/ti/am335x/README | 53 ++-- drivers/mtd/nand/omap_gpmc.c | 565 +++++++++++++-------------------------- 3 files changed, 217 insertions(+), 408 deletions(-)
Signature · 3 lines
Ezequiel Garc?a, Free Electrons Embedded Linux, Kernel and Android Engineering http://free-electrons.com
Gupta
v9 cover mtd: nand: omap: optimize OMAP NAND driver
In reply to Pekon GuptaHello Scott,
Quoted 22 lines
From: Gupta, Pekon
Subject: [PATCH v9 0/4] mtd: nand: omap: optimize OMAP NAND driverOMAP NAND drivers support various ecc-schemes like (HAM1, BCH4, BCH8..) And, each ecc-scheme usually has 2 implementations
-HAM1_HWorBCHx_HW(using in-build h/w engine for all computation)
-HAM1_SWorBCHx_HW_DETECTION_SW) (using software library for some computation)This patch series is Part-2 (split) from earlier series, as mentioned below:
http://lists.denx.de/pipermail/u-boot/2013-November/167270.htmlThis series aims at following:
- cleans-up redundant code, duplicate in different ecc-schemes
- optimizes ecc calculation and correction paths
- improves scalability and readability, for adding newer ecc-schemes in future.
- updates following callbacks for different ecc-schmes:
[PATCH 1/4]nand_chip->ecc.hwctl(): configures h/w controller before for Read/Write acccess [PATCH 2/4]nand_chip->ecc.calculate(): calculates and fetches ECC syndrome from h/w controller. [PATCH 3/4]nand_chip->ecc.correct(): detects and corrects ECC errors on read access [PATCH 4/4] update doc and README
Quoted 3 lines
*changes in v9*
- split from main series, rebased and forked independently
Request you to please review this series. And accept it if it looks okay.
And a bug-fix which causes regression for x8 NAND devices using HAM1 ecc-scheme. This was reported & (un-officially) tested by Stefan Roese <[email protected]>.
with regards, pekon
v9 cover mtd: nand: omap: optimize OMAP NAND driver
Quoted 26 lines · Pekon
Quoted 22 lines
From: Gupta, Pekon
Subject: [PATCH v9 0/4] mtd: nand: omap: optimize OMAP NAND driverOMAP NAND drivers support various ecc-schemes like (HAM1, BCH4, BCH8..) And, each ecc-scheme usually has 2 implementations
-HAM1_HWorBCHx_HW(using in-build h/w engine for all computation)
-HAM1_SWorBCHx_HW_DETECTION_SW) (using software library for some computation)This patch series is Part-2 (split) from earlier series, as mentioned below:
http://lists.denx.de/pipermail/u-boot/2013-November/167270.htmlThis series aims at following:
- cleans-up redundant code, duplicate in different ecc-schemes
- optimizes ecc calculation and correction paths
- improves scalability and readability, for adding newer ecc-schemes in future.
- updates following callbacks for different ecc-schmes:
[PATCH 1/4]nand_chip->ecc.hwctl(): configures h/w controller before for Read/Write acccess [PATCH 2/4]nand_chip->ecc.calculate(): calculates and fetches ECC syndrome from h/w controller. [PATCH 3/4]nand_chip->ecc.correct(): detects and corrects ECC errors on read access [PATCH 4/4] update doc and READMEQuoted 3 lines
*changes in v9*
- split from main series, rebased and forked independently
Quoted 5 lines
Request you to please review this series. And accept it if it looks okay.
And a bug-fix which causes regression for x8 NAND devices using HAM1 ecc-scheme. This was reported & (un-officially) tested by Stefan Roese <[email protected]>.
I tested this patch series plus the fix from Pekon on an x8 and x16 OMAP3530 board. Both without any problems. So for this series:
Thanks,
Stefan