• State
  • Locked Locked
  • Replies 4 replies
  • Subscribers 98 subscribers
  • Views 5138 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

skip bad blocks in uboot & kernel

zanget
Expert 2150 points

I’m trying to add the new NAND recently to our custom board in DM365 with DVSDK(dvsdk_dm365-evm_4_01_00_09_setuplinux)

 After patch with the following clip,

http://processors.wiki.ti.com/index.php/DM365_Nand_ECC_layout

 

The kernel seems cannot skip the bad block and we get problem with :

 

cat start.sh

start.sh: Input/output error

 

 

 

I get the error form u-boot when it’s starting:

 

 

U-Boot 2010.12-rc2 (Jun 15 2011 - 10:29:05)

 

Cores: ARM 297 MHz

DDR:   243 MHz

I2C:   ready

DRAM:  128 MiB

NAND:  128 MiB

MMC:   davinci: 0, davinci: 1

Bad block table found at page 65472, version 0x01

Bad block table found at page 65408, version 0x01

nand_bbt: ECC error while reading bad block table

nand_read_bbt: Bad block at 0x000002e60000

nand_read_bbt: Bad block at 0x000003b40000

nand_read_bbt: Bad block at 0x000007340000

 

this is what I get so far :

NAND:  K9F1G08U0D  128M x 8 Bit NAND Flash Memory

• Organization

- Memory Cell Array : (128M + 4M) x 8bit

- Data Register : (2K + 64) x 8bit

• Automatic Program and Erase

- Page Program : (2K + 64)Byte

- Block Erase : (128K + 4K)Byte

 

Linux:    2.6.32.17-psp03.01.01.38

u-boot:      2010.12-rc1-psp03.01.01.38

 

u-boot NAND configuration:

 

#define CONFIG_NAND_DAVINCI

 

#define CONFIG_SYS_NAND_CS                2

#define CONFIG_SYS_NAND_USE_FLASH_BBT

#define CONFIG_SYS_NAND_4BIT_HW_ECC_OOBFIRST

#define CONFIG_SYS_NAND_PAGE_2K

 

#define CONFIG_SYS_NAND_LARGEPAGE

/* Confirmed: 0x02000000 ASYNC EMIF Data (CE0) address see page 16 of device-specific data manual */

#define CONFIG_SYS_NAND_BASE_LIST          { 0x02000000, }

/* socket has two chipselects, nCE0 gated by address BIT(14) */

#define CONFIG_SYS_MAX_NAND_DEVICE    1

#define CONFIG_SYS_NAND_MAX_CHIPS       1

 

 

Changes in the board-dm365.c:

 

//mike added to support the hw ecc for dm365 uboot/ubl

static struct nand_ecclayout dm365_evm_nand_ecclayout = {

         .eccbytes = 40,

         .eccpos              = {6, 7, 8, 9, 10, 11, 12, 13, 14, 15,

                              22, 23, 24, 25, 26, 27, 28, 29, 30, 31,

                              38, 39, 40, 41, 42, 43, 44, 45, 46, 47,

                              54, 55, 56, 57, 58, 59, 60, 61, 62, 63,

         },

         .oobfree   = {{2, 4}, {16, 6}, {32, 6}, {48, 6} },

};

 

 

 

static struct davinci_nand_pdata davinci_nand_data = {

    .mask_chipsel           = 0,    // only one nand see below

         .parts                          = davinci_nand_partitions,

         .nr_parts           = ARRAY_SIZE(davinci_nand_partitions),

    .ecc_mode                 = NAND_ECC_HW,    // no check ecc  : NAND_ECC_NONE

         .options             = NAND_USE_FLASH_BBT,

         .ecc_bits           = 4,

//mike : added

         .ecclayout                  = &dm365_evm_nand_ecclayout,

};

 

If we use

         .mask_chipsel                   = BIT(14),

The kernel keep to tell us that :

 

NAND device: Manufacturer ID: 0xec, Chip ID: 0xf1 (Samsung NAND 128MiB 3,3V 8-bit)

2 NAND chips detected

nand_bbt: ECC error while reading bad block table

nand_bbt: ECC error while reading bad block table

Creating 8 MTD partitions on "davinci_nand.0":

 

More info when start:

List of all partitions:

1f00            4096 mtdblock0 (driver?)

1f01            1024 mtdblock1 (driver?)

1f02            3072 mtdblock2 (driver?)

1f03           20480 mtdblock3 (driver?)

1f04            1024 mtdblock4 (driver?)

1f05           32768 mtdblock5 (driver?)

1f06            4096 mtdblock6 (driver?)

1f07          195584 mtdblock7 (driver?)

No filesystem could mount root, tried:  cramfs

Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(31,3)

Backtrace:

 [<c008c5f4>] (dump_backtrace+0x0/0x114) from [<c039d280>] (dump_stack+0x18/0x1c)

r7:00008000 r6:c2015000 r5:c0026808 r4:c04ce3f8

 [<c039d268>] (dump_stack+0x0/0x1c) from [<c039d2d8>] (panic+0x54/0x12c)

 [<c039d284>] (panic+0x0/0x12c) from [<c0008fbc>] (mount_block_root+0x1e0/0x220)

r3:00000000 r2:00000020 r1:c2029f60 r0:c0435da6

 [<c0008ddc>] (mount_block_root+0x0/0x220) from [<c00090c0>] (mount_root+0xc4/0xfc)

r8:00000000 r7:00000000 r6:00000000 r5:c0026808 r4:01f00003

 [<c0008ffc>] (mount_root+0x0/0xfc) from [<c0009268>] (prepare_namespace+0x170/0x1c8)

r5:c0026808 r4:c04cdf20

 [<c00090f8>] (prepare_namespace+0x0/0x1c8) from [<c00084bc>] (kernel_init+0xe4/0x118)

r5:00000000 r4:c04cdce0

[<c00083d8>] (kernel_init+0x0/0x118) from [<c00a0d88>] (do_exit+0x0/0x668)

r5:00000000 r4:00000000

 

 

 

we tried to flash the whole chip and write the data to the mtdblock5 list below:

 

nand write 0x80700000 0x1d00000 0x2000000

 

NAND write: device 0 offset 0x1d00000, size 0x2000000

Skip bad block 0x02e60000

Skip bad block 0x03b40000

 33554432 bytes written: OK

 

But the kernel cannot skip the bad block we found it is bad in uboot:

 

NAND device: Manufacturer ID: 0xec, Chip ID: 0xf1 (Samsung NAND 128MiB 3,3V 8-bit)

nand_bbt: ECC error while reading bad block table

Creating 8 MTD partitions on "davinci_nand.0":

0x000000000000-0x000000400000 : "bootloader"

 

When MTD BLOCK5 is mounted, we tried to do something in that mtd-part, we got:

cat start.sh

start.sh: Input/output error

 

we also tried to flash the app in kernel without luck:

/ # flash_eraseall  /dev/mtd5

Erasing 128 Kibyte @ 1140000 -- 53 % complete.

Skipping bad block at 0x01160000

Erasing 128 Kibyte @ 1e20000 -- 94 % complete.

Skipping bad block at 0x01e40000

Erasing 128 Kibyte @ 2000000 -- 100 % complete

 

/ # nandwrite /dev/mtd5 app.img

Writing data to block 0 at offset 0x0

Writing data to block 1 at offset 0x20000

Writing data to block 2 at offset 0x40000

<Snap>

Writing data to block 137 at offset 0x1120000

Writing data to block 138 at offset 0x1140000

Writing data to block 139 at offset 0x1160000

Bad block at 1160000, 1 block(s) from 1160000 will be skipped

Writing data to block 140 at offset 0x1180000

Writing data to block 141 at offset 0x11a0000

Writing data to block 142 at offset 0x11c0000

<Snap>

Writing data to block 164 at offset 0x1480000

Writing data to block 165 at offset 0x14a0000

Writing data to block 166 at offset 0x14c0000

 

Don’t know why the kernel found just one bad block when write date into that mtd-part.

 

After reboot, u-boot seems to print the same thing with the previous steps we used u-boot to flash the app:

 

Bad block table found at page 65472, version 0x01

Bad block table found at page 65408, version 0x01

nand_bbt: ECC error while reading bad block table

nand_read_bbt: Bad block at 0x000002e60000

nand_read_bbt: Bad block at 0x000003b40000

nand_read_bbt: Bad block at 0x000007340000

 

And the dmesg gives the same message comp to the previous steps:

 

NAND device: Manufacturer ID: 0xec, Chip ID: 0xf1 (Samsung NAND 128MiB 3,3V 8-bit)

nand_bbt: ECC error while reading bad block table

Creating 8 MTD partitions on "davinci_nand.0":

0x000000000000-0x000000400000 : "bootloader"

0x000000400000-0x000000500000 : "params"

0x000000500000-0x000000800000 : "kernel"

0x000000800000-0x000001c00000 : "rootfs"

0x000001c00000-0x000001d00000 : "driver"

0x000001d00000-0x000003d00000 : "app"

0x000003d00000-0x000004100000 : "config"

0x000004100000-0x000008000000 : "ext"

davinci_nand davinci_nand.0: controller rev. 2.3

 

Any ideas? Any idea will be appreciate.

Thinks in advance.

Regards, Mike.

 

  • 0
    Expert 2150 points

    Any comments would be sincerely appreciated...

  • 0 in reply to zanget
    TI__Expert 5170 points

    Hi,

    I am not sure what is the problem here. Are you worried about "ECC error while reading bad block table" message? This is just a warning message which can be ignored. I assume this error may go if you define CONFIG_SYS_NAND_USE_FLASH_BBT in the configs file.

    If you are worried that, you are not able to see the file present in /dev/mtd5, make sure that you erase the partition, store the file and then reboot the board and see whether the file exists and then read from it.

    Regards, Sudhakar

  • 0 in reply to Sudhakar Rajashekhara
    Expert 2150 points

    hi, Sudhakar.

    Thinks for your post.

    we had

    #define CONFIG_SYS_NAND_USE_FLASH_BBT

    in uboot but we still encounter the same Warning message .

    Both u-boot and kernel report the same message to us so we dont konw what to do next if we want to use bbt.

    we had even tried to erase the whole chip and then re-write ubl+u-boot to the NAND flash but the message does not disappear.

    the significant problem is that we found we faced with bad block that need to be skipped form the kernel but it turns out it cannot.

    we found that the problem comes form the cramfs:

    cramfs works very nicely on the /dev/mtdblockX device and does all I
    want. But of course cramfs is usable only as long as the NAND has no
    bad blocks since cramfs itself is not flash-aware. Throwing away the
    hardware whenever there's a block error would generate unacceptable
    levels of waste and repair costs in this project.

    we are now try to patch the carmfs to skip the bad blocks...

    maybe it is a good idea to try with other fs.. we dont know.

    thinks in advance.

    regards, Mike

  • 0 in reply to zanget
    TI__Expert 5170 points

    Hi,

    You can consider using JFFS2 and YAFFS2 file systems which are specifically designed for MTD devices. Out of these two, support for JFFS2 is already present in the kernel but YAFFS2 support is available as a patch.

    Regards, Sudhakar