ECC Error when using H27UBG8T2BTR/CTR with Linux/iMX28

cuigao · ‎05-19-2014

Has anybody tried H27UBG8T2BTR/CTR on linux for iMX28?

I'm trying to add some new NAND flash support to the kernel for iMX28 from L2.6.35_10.12.01_SDK. But it seems not to work. I'm using UBIFS as the filesystem.

For the NAND H27UBG8T2CTR, here is the information printed by kernel while booting. I added some items in the array in nand_device_info.c, and modified code to support the page geometry of 8192+640.

-----------------------------

NAND Flash Device Information

-----------------------------

Manufacturer : Hynix (0xad)

Device Code : 0xd7

Cell Technology : MLC

Chip Size : 4 GiB

Pages per Block : 256

Page Geometry : 8192+640

ECC Strength : 20 bits

ECC Size : 1024 B

Data Setup Time : 15 ns

Data Hold Time : 10 ns

Address Setup Time: 20 ns

GPMI Sample Delay : 6 ns

tREA : 20 ns

tRLOH : 5 ns

tRHOH : 15 ns

Description : H27UBG8T2BTR (4GB, 1CE); H27UBG8T2CTR (4GB, 1CE)

-----------------

Physical Geometry

-----------------

Chip Count : 1

Page Data Size in Bytes: 8192 (0x2000)

Page OOB Size in Bytes : 640

Block Size in Bytes : 2097152 (0x200000)

Block Size in Pages : 256 (0x100)

Chip Size in Bytes : 4294967296 (0x100000000)

Chip Size in Pages : 524288 (0x80000)

Chip Size in Blocks : 2048 (0x800)

Medium Size in Bytes : 4294967296 (0x100000000)

------------

NFC Geometry

------------

ECC Algorithm : BCH

ECC Strength : 20

Page Size in Bytes : 8832

Metadata Size in Bytes : 10

ECC Chunk Size in Bytes: 512

ECC Chunk Count : 16

Payload Size in Bytes : 8192

Auxiliary Size in Bytes: 28

Auxiliary Status Offset: 12

Block Mark Byte Offset : 7694

Block Mark Bit Offset : 4

When I created a new ubi volume on /dev/mtd1 and mounted it at the first time, evrything is ok. And after I umounted it and then mount it again, I got some io error. The error -74 is EBADMSG which means ECC failed. See the following logs:

$ ubiformat /dev/mtd1

ubiformat: mtd1 (nand), size 2013265920 bytes (1.9 GiB), 960 eraseblocks of 2097

152 bytes (2.0 MiB), min. I/O size 8192 bytes

libscan: scanning eraseblock 959 -- 100 % complete

ubiformat: 959 eraseblocks are supposedly empty

ubiformat: 1 bad eraseblocks found, numbers: 517

ubiformat: formatting eraseblock 959 -- 100 % complete

$

$ ubiattach /dev/ubi_ctrl -m 1

UBI: attaching mtd1 to ubi0

UBI: physical eraseblock size: 2097152 bytes (2048 KiB)

UBI: logical eraseblock size: 2080768 bytes

UBI: smallest flash I/O unit: 8192

UBI: VID header offset: 8192 (aligned 8192)

UBI: data offset: 16384

UBI: attached mtd1 to ubi0

UBI: MTD device name: "gpmi-nfc-general-use"

UBI: MTD device size: 1920 MiB

UBI: number of good PEBs: 959

UBI: number of bad PEBs: 1

UBI: max. allowed volumes: 128

UBI: wear-leveling threshold: 4096

UBI: number of internal volumes: 1

UBI: number of user volumes: 0

UBI: available PEBs: 946

UBI: total number of reserved PEBs: 13

UBI: number of PEBs reserved for bad PEB handling: 9

UBI: max/mean erase counter: 0/0

UBI: image sequence number: 440121938

UBI: background thread "ubi_bgt0d" started, PID 2832

UBI device number 0, total 959 LEBs (1995456512 bytes, 1.9 GiB), available 946 L

EBs (1968406528 bytes, 1.8 GiB), LEB size 2080768 bytes (2.0 MiB)

$

$ ubimkvol /dev/ubi0 -N rootfs -s 1024MiB

ubiblka: unknown partition table

Volume ID 0, size 517 LEBs (1075757056 bytes, 1.0 GiB), LEB size 2080768 bytes (

2.0 MiB), dynamic, name "rootfs", alignment 1

$

$ mount -t ubifs ubi0:rootfs /mnt/pp/

UBIFS: default file-system created

UBIFS: mounted UBI device 0, volume 0, name "rootfs"

UBIFS: file system size: 1057030144 bytes (1032256 KiB, 1008 MiB, 508 LEBs)

UBIFS: journal size: 33292288 bytes (32512 KiB, 31 MiB, 16 LEBs)

UBIFS: media format: w4/r0 (latest is w4/r0)

UBIFS: default compressor: lzo

UBIFS: reserved for root: 4952683 bytes (4836 KiB)

$

$ umount /mnt/pp/

UBIFS: un-mount UBI device 0, volume 0

$

$ mount -t ubifs ubi0:rootfs /mnt/pp/

UBI error: ubi_io_read: error -74 while reading 2080768 bytes from PEB 513:16384

, read 2080768 bytes

UBI error: ubi_io_read: error -74 while reading 2080768 bytes from PEB 513:16384

, read 2080768 bytes

UBI error: ubi_io_read: error -74 while reading 2080768 bytes from PEB 514:16384

, read 2080768 bytes

UBIFS: recovered master node from LEB 1

UBIFS: mounted UBI device 0, volume 0, name "rootfs"

UBIFS: file system size: 1057030144 bytes (1032256 KiB, 1008 MiB, 508 LEBs)

UBIFS: journal size: 33292288 bytes (32512 KiB, 31 MiB, 16 LEBs)

UBIFS: media format: w4/r0 (latest is w4/r0)

UBIFS: default compressor: lzo

UBIFS: reserved for root: 4952683 bytes (4836 KiB)

$

$ umount /mnt/pp/

UBIFS: un-mount UBI device 0, volume 0

But if I run nandtest with /dev/mtd1, it could pass it without errors. I tried ext3 filesystem on /dev/mtdblock1, it also failed with some I/O error when I was writing a lot of files on it.

And I tried another NAND H27UBG8T2BTR. When I was trying to mount the UBI partition the second time, it told me some PEB is scrubbed and run torture test for it.

So could iMX28 support such kind of NAND? Or is anything I did wrong? Thanks!

The datasheet could be found in the attachment.

igorpadykov · ‎05-21-2014

Hi Gao

sorry to say you but 8K NANDs are not supported by i.MX28

proccesors and Freescale did not test them. You should choose

4K page NANDs, it is supported.

Best regards

chip

-----------------------------------------------------------------------------------------------------------------------

Note: If this post answers your question, please click the Correct Answer button. Thank you!

-----------------------------------------------------------------------------------------------------------------------

cuigao · ‎05-25-2014

Hi chipexpert,

Thanks for your reply. But it is getting hard to buy the 4K page NANDs nowadays. 8K/16K page NANDs are much more common and cheaper. That's why I'm trying to use them. I've ported another 8K page NAND H27UAG8T2BTR one year ago, and it works well. But it becomes too expensive now.

Could you explain some reason why iMX28 does not support 8K NANDs? I saw the registers have enough addressing bits to support 8K page NANDs. They are all 16 bits registers to hold a page address. So I can't understand why it fails.

And does freescale have a plan to support with new NANDs of 8K/16K pages?

Thank you!

igorpadykov · ‎05-25-2014

Hi Gao

not, sorry there are no plans to support it.

Official Freescale BSPs do not support them.

Best regards

chip

ECC Error when using H27UBG8T2BTR/CTR with Linux/iMX28

ECC Error when using H27UBG8T2BTR/CTR with Linux/iMX28

i.MX2x

i.MX50

i.MX51

i.MX53

i.MX6_All

i.MX6DL

i.MX6Dual

i.MX6Quad

i.MX6S

i.MX6SL

Linux

Suspected Software Defect