topic Re: NAND ecc error in T1022 in QorIQ

NAND ecc error in T1022

Kernelhacker — Mon, 02 Mar 2020 09:27:55 GMT

Custom board based on T1022.

Nand flash: MT29F32G08ABAAAWP-Z:A

Page size = 8192; Block size = 1024KiB; Block count = 4096, Memory size = 4096MiB, 8-bit ECC

filesystem: ubifs used as root

The file system mounts properly without any error. But on certain occasions (randomly) we are getting these errors:

UBIFS error (ubi0:0 pid 1881): try_read_node: cannot read node type 1 from LEB 17:803080, error -74
fsl,ifc-nand fff800000.nand: NAND Flash ECC Uncorrectable Error
ubi0 warning: ubi_io_read: error -74 (ECC error) while reading 2590 bytes from PEB 19:819464, read only 2590 bytes, retry
fsl,ifc-nand fff800000.nand: NAND Flash ECC Uncorrectable Error
ubi0 warning: ubi_io_read: error -74 (ECC error) while reading 2590 bytes from PEB 19:819464, read only 2590 bytes, retry
fsl,ifc-nand fff800000.nand: NAND Flash ECC Uncorrectable Error
ubi0 warning: ubi_io_read: error -74 (ECC error) while reading 2590 bytes from PEB 19:819464, read only 2590 bytes, retry
fsl,ifc-nand fff800000.nand: NAND Flash ECC Uncorrectable Error
ubi0 error: ubi_io_read: error -74 (ECC error) while reading 2590 bytes from PEB 19:819464, read 2590 bytes
CPU: 1 PID: 1881 Comm: ls Tainted: P O 4.1.8-rt8 #19
Call Trace:
[e64f9a50] [c0676320] dump_stack+0x78/0xa0 (unreliable)
[e64f9a60] [c03af988] ubi_io_read+0x218/0x400@@@@#@
[e64f9ac0] [c03acdb0] ubi_eba_read_leb+0x150/0x480
[e64f9b10] [c03abe80] ubi_leb_read+0xb0/0x150
[e64f9b40] [c022cc30] ubifs_leb_read+0x30/0xa0
[e64f9b70] [c022ed68] ubifs_read_node+0xa8/0x380
[e64f9ba0] [c0252a50] ubifs_tnc_read_node+0x190/0x200
[e64f9c00] [c023338c] ubifs_tnc_locate+0x1cc/0x210
[e64f9c50] [c0222a70] do_readpage+0x1e0/0x5d0
[e64f9cb0] [c022412c] ubifs_readpage+0x4c/0x490
[e64f9d10] [c00c33c8] filemap_fault+0x2f8/0x4b0
[e64f9d50] [c00e60f4] __do_fault+0x44/0xc0
[e64f9d80] [c00e9ffc] handle_mm_fault+0xa7c/0xf00
[e64f9e00] [c0016a94] do_page_fault+0x2e4/0x620
[e64f9e40] [c000fb1c] handle_page_fault+0xc/0x80
--- interrupt: 301 at do_page_fault+0x3a0/0x620
LR = handle_page_fault+0xc/0x80
[e64f9f00] [00000000] (null) (unreliable)
[e64f9f40] [c000fb1c] handle_page_fault+0xc/0x80
--- interrupt: 401 at 0x1000b320
LR = 0x10002f30
UBIFS error (ubi0:0 pid 1881): ubifs_read_node: bad node type (55 but expected 1)
UBIFS error (ubi0:0 pid 1881): ubifs_read_node: bad node at LEB 17:803080, LEB mapping status 1
Not a node, first 24 bytes:
00000000: 37 3f 37 37 bf bf bf 37 37 3f 37 37 37 37 37 37 3f 3f 37 37 37 37 37 37 7?77...77?777777??777777
CPU: 1 PID: 1881 Comm: ls Tainted: P O 4.1.8-rt8 #19
Call Trace:
[e64f9b60] [c0676320] dump_stack+0x78/0xa0 (unreliable)
[e64f9b70] [c022efb8] ubifs_read_node+0x2f8/0x380

1. nanddump of the first few pages of the partition we see that they are all 0x37 or 3f or 0xbf and like wise.

2. On next boot-up there are no errors and filesystem mounts without any error. Nanddump show valid pages.

3. We are not certain what causes this error.

4. From the error and nanddump it seems, that since all the data are corrupted, ECC uncorrectable error is coming on nand read.

PFA schematic, pls give your suggestions.

Re: NAND ecc error in T1022

ufedor — Mon, 02 Mar 2020 10:53:41 GMT

1) Which NXP Linux SDK is used?

2) What is the value of the IFC_CSOR0_NAND?

Re: NAND ecc error in T1022

Kernelhacker — Tue, 03 Mar 2020 06:17:11 GMT

1) Which NXP Linux SDK is used?

Linux SDK v2.0

2) What is the value of the IFC_CSOR0_NAND?

0xFFE124130 = 0x9518c200

Re: NAND ecc error in T1022

ufedor — Tue, 03 Mar 2020 06:24:51 GMT

How many boards were tested?

Re: NAND ecc error in T1022

Kernelhacker — Tue, 03 Mar 2020 06:33:05 GMT

3-4 boards. all have this problem

Re: NAND ecc error in T1022

ufedor — Tue, 03 Mar 2020 06:38:54 GMT

Please try the IFC_CSOR0_NAND = 0xA518c200

Re: NAND ecc error in T1022

Kernelhacker — Tue, 03 Mar 2020 08:23:46 GMT

Unable to write NAND flash with:

IFC_CSOR0_NAND = 0xA518c200

=> run flash_nand_dtb
Using FM1@DTSEC4 device
TFTP from server 192.168.1.10; our IP address is 192.168.1.100
Filename 'dtb_oxph'.
Load address: 0x1000000
Loading: ###
1.4 MiB/s
done
Bytes transferred = 33245 (81dd hex)

NAND erase: device 0 offset 0x1100000, size 0x100000
Erasing at 0x1100000 -- 100% complete.
OK

NAND write: device 0 offset 0x1100000, size 0x81dd
NAND write to offset 1100000 failed -5
0 bytes written: ERROR

=> md.l 0xfe124130 1
fe124130: a518c200

Re: NAND ecc error in T1022

ufedor — Tue, 03 Mar 2020 08:28:46 GMT

Have you reconfigured and re-built U-Boot so 24-bit BCH is used?

Re: NAND ecc error in T1022

Kernelhacker — Tue, 03 Mar 2020 08:32:23 GMT

yes

Re: NAND ecc error in T1022

Kernelhacker — Wed, 04 Mar 2020 04:07:10 GMT

Could it be a hardware issue?

Is the schematic ok?

Re: NAND ecc error in T1022

ufedor — Wed, 04 Mar 2020 04:13:08 GMT

Please use a digital scope to check signals behaviour at both sides of the level translators U13-U16.

Re: NAND ecc error in T1022

Kernelhacker — Thu, 30 Apr 2020 04:31:33 GMT

the level translator where not behaving properly.

Solved now

Thanks