Uncorrectable ECC error

Hi,

if I do a blkid, I do get these errors

root@nws:~# blkid
[18327.978487] __nand_correct_data: uncorrectable ECC error
[18327.983891] __nand_correct_data: uncorrectable ECC error
[18327.989228] blk_update_request: I/O error, dev mtdblock5, sector 0
[18327.996072] __nand_correct_data: uncorrectable ECC error
[18328.002917] __nand_correct_data: uncorrectable ECC error
[18328.008257] blk_update_request: I/O error, dev mtdblock5, sector 0
[18328.014463] Buffer I/O error on dev mtdblock5, logical block 0, async page read
[18328.109550] __nand_correct_data: uncorrectable ECC error
[18328.114921] __nand_correct_data: uncorrectable ECC error
[18328.120988] blk_update_request: I/O error, dev mtdblock6, sector 512
[18328.127604] __nand_correct_data: uncorrectable ECC error
[18328.133624] __nand_correct_data: uncorrectable ECC error
[18328.138960] blk_update_request: I/O error, dev mtdblock6, sector 512
[18328.145340] Buffer I/O error on dev mtdblock6, logical block 64, async page read
/dev/sda1: LABEL="rootfs" UUID="9fa5b89f-03c0-4fdf-a026-860cf6c673de" TYPE="ext4" PARTUUID="cc74c377-01"
root@nws:~#

What's wrong here? Is this EMC troubles maybe?



Edited 1 time(s). Last edit at 01/25/2018 07:13AM by lille.

lille,

> What's wrong here? Is this EMC troubles maybe?

There is nothing wrong. blkid always spits out error like that if the mtdblock devices are not defined. If you do fdisk, you will see that they are not. Which kernel are you running?

uname -a

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

thank you for the clarification. i was assuming not seeing this 'ecc error' message before when i looked at the dmesg output. maybe I'm just wrong with this assumption.

[Thu Jan 25 09:48:25 2018] __nand_correct_data: uncorrectable ECC error
[Thu Jan 25 09:48:25 2018] __nand_correct_data: uncorrectable ECC error
[Thu Jan 25 09:48:25 2018] blk_update_request: I/O error, dev mtdblock5, sector 0
[Thu Jan 25 09:48:25 2018] __nand_correct_data: uncorrectable ECC error
[Thu Jan 25 09:48:25 2018] __nand_correct_data: uncorrectable ECC error
[Thu Jan 25 09:48:25 2018] blk_update_request: I/O error, dev mtdblock5, sector 0
[Thu Jan 25 09:48:25 2018] Buffer I/O error on dev mtdblock5, logical block 0, async page read

my machine is on kernel

root@nws:~# uname -a
Linux nws 4.12.1-kirkwood-tld-1 #1 PREEMPT Sat Jul 15 21:40:50 PDT 2017 armv5tel GNU/Linux

root@nws:~# fdisk -l
Disk /dev/mtdblock0: 1 MiB, 1048576 bytes, 2048 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size [  163.783886] __nand_correct_data: uncorrectable ECC error
(minimum/optimal[  163.789431] __nand_correct_data: uncorrectable ECC error
): 512 bytes / 5[  163.796074] blk_update_request: I/O error, dev mtdblock5, sector 0
12 bytes


Disk /dev/mtdblock[  163.805053] __nand_correct_data: uncorrectable ECC error
1: 512 KiB, 5242[  163.811881] __nand_correct_data: uncorrectable ECC error
88 bytes, 1024 s[  163.818578] blk_update_request: I/O error, dev mtdblock5, sector 0
[  163.826166] Buffer I/O error on dev mtdblock5, logical block 0, async page read
ectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes


Disk /dev/mtdblock2: 512 KiB, 524288 bytes, 1024 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes


Disk /dev/mtdblock3: 512 KiB, 524288 bytes, 1024 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes


Disk /dev/mtdblock4: 10 MiB, 10485760 bytes, 20480 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes




Disk /dev/mtdblock6: 47.8 MiB, 50069504 bytes, 97792 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes


Disk /dev/mtdblock7: 10 MiB, 10485760 bytes, 20480 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes


Disk /dev/mtdblock8: 47.8 MiB, 50069504 bytes, 97792 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes

lille,

Ah. I was expected to see undefined mtd block devices, where blkid would spit out errors no matter what.

However, this kernel version has mtd block device built in. So the other mtds are OK, but not mtd5. That's not very good, it is a real problem.

Which box is this? please post the entire log of

dmesg
cat /proc/mtd

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)



Edited 1 time(s). Last edit at 01/25/2018 02:28PM by bodhi.

bodhi, it's a ZyXEL NSA325 v2

root@cat /proc/mtd
dev:    size   erasesize  name
mtd0: 00100000 00020000 "uboot"
mtd1: 00080000 00020000 "uboot_env"
mtd2: 00080000 00020000 "key_store"
mtd3: 00080000 00020000 "info"
mtd4: 00a00000 00020000 "etc"
mtd5: 00a00000 00020000 "kernel_1"
mtd6: 02fc0000 00020000 "rootfs1"
mtd7: 00a00000 00020000 "kernel_2"
mtd8: 02fc0000 00020000 "rootfs2"
root@nws:~#

this box did not run 3 weeks in total, you could call it new :( wtf

lille,

> this box did not run 3 weeks in total, you could
> call it new :( wtf

You could obmit that part of NAND in mtd definition like I do. If you don't plan to go back to stock, then no worry. Just define it like this. Note that the second partition is not used either, I included it for illustration purpose.

fw_setenv mtdparts 'mtdparts=orion_nand:0x100000(uboot),0x80000(uboot_env)'

The result is

fw_printenv mtdparts

mtdparts=mtdparts=orion_nand:0x100000(uboot),0x80000(uboot_env)

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

I see exactly the same error on my NSA325v2

[  160.603907] __nand_correct_data: uncorrectable ECC error
[  160.609263] __nand_correct_data: uncorrectable ECC error
[  160.623771] print_req_error: I/O error, dev mtdblock5, sector 0
[  160.634636] __nand_correct_data: uncorrectable ECC error
[  160.639977] __nand_correct_data: uncorrectable ECC error
[  160.663770] print_req_error: I/O error, dev mtdblock5, sector 0
[  160.669726] Buffer I/O error on dev mtdblock5, logical block 0, async page read

Is it safe to ignore or is there any way to fix it? I not plan to go back to stock again.

sebr,

> I see exactly the same error on my NSA325v2
>

> 
> [  160.603907] __nand_correct_data: uncorrectable
> ECC error
> [  160.609263] __nand_correct_data: uncorrectable
> ECC error
> [  160.623771] print_req_error: I/O error, dev
> mtdblock5, sector 0
> [  160.634636] __nand_correct_data: uncorrectable
> ECC error
> [  160.639977] __nand_correct_data: uncorrectable
> ECC error
> [  160.663770] print_req_error: I/O error, dev
> mtdblock5, sector 0
> [  160.669726] Buffer I/O error on dev mtdblock5,
> logical block 0, async page read
> 
>

> Is it safe to ignore or is there any way to fix
> it? I not plan to go back to stock again.

This error could occur if you executing some command such as blkid, and in that case it is perfectly normal. If you keep seeing it without doing anything that tries to read the mtd partitions then it is not good.

However, I've seen this error bfore on my box too. What I did was to make the box ignoring it altogether by redefine the mtdparts:

fw_setenv mtdparts 'mtdparts=orion_nand:0x100000(uboot),0x80000(uboot_env)'

So I don't care to see the rest of the mtds except for u-boot and u-boot envs region. I could try to fix it, but thinking I would never need to boot back to stock, so chose to not worry about it now.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)



Edited 1 time(s). Last edit at 09/17/2018 05:53PM by bodhi.

I get this a lot. I recently messed with my server to consolidate drives and backup the rootfs and noticed this is happening every boot. It doesn't happen when I run

blkid

but it happens when I attach a USB device. I get:

[156624.656218] __nand_correct_data: uncorrectable ECC error
[156624.661674] blk_update_request: I/O error, dev mtdblock0, sector 2040
[156624.668905] __nand_correct_data: uncorrectable ECC error
[156624.674371] blk_update_request: I/O error, dev mtdblock0, sector 2040
[156624.680929] Buffer I/O error on dev mtdblock0, logical block 255, async page read

It seems to not be much of an issue, but still I'd like to eliminate it. My setup uses ubifs in order to provide uImage and uInitrd. I then mount it onto /boot so that flashing the kernel updates them. (for reasons I won't go into, my boot drive cannot host uImage/uInitrd. I utilize NAND for that, but tftp works fine also)

my mtdparts are defined as such:

mtdparts=orion_nand:1M(u-boot),-(rootfs)

and my fw_foo commands work fine using 0xc0000 offset and 0x20000 size in /etc/fw_env.config

Is it just a matter of including uboot-env part, or something else? Should I just not even be concerned since there are no apparent issues?

Nematocyst,

It is not a real issue (except when it occurs too often and you want to get rid off this spamming your sys log). Just a matter of your Linux block devices (mtd0 is a block device) driver don't recognize correctly what's in mtd0. The box bootROM access mtd0 without problem (e.g. it can load and start u-boot on mtd0).

But the fact that when you attach USB drive and see these messages, your rootfs must be different from what I released for Kirkwood boxes. Which kernel, and rootfs are you running on your system?

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

It's been a while since I installed it. but it's basically a straight debian stretch install. I d/l the armv5 uInitd and uInitrd, tftp booted, and ran the installer using serial console, iirc. It's currently running 4.9.0-8-marvell (4.9.130-2, 2018-10-27). and ofc, rootfs was generated during install. this is the same install that I had issues running out of memory a year or so ago that caused me to remove systemd.

I certainly don't need the most up to date kernel. It's strictly a headless home server. apache, exim4, dovecot, inadyn, samba, cups, tftp



Edited 1 time(s). Last edit at 01/02/2019 10:34PM by Nematocyst.

Nematocyst,

> basically a straight debian stretch install.

> currently running 4.9.0-8-marvell (4.9.130-2,
> 2018-10-27).

Ah, so that explained it. Mainline kernel configuration might not have all needed modules and perhaps udev rules to deal with all types of block devices. The system certainly correct coming from mainline, but some improvement needed to deal with all types of configurations, even some simple as yours.

Kernel 4.9.x is fine. I suspect it is just some more configurations are needed.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

Another thought,

You might want to use the DTB that I released for your box (which box is this?) to see if it will make any difference.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

Oh it's a dockstar, sorry i didn't mention that. I've got 2 dockstars and 1 sheevaplug. I keep thinking if something goes wrong, I've got an immediate redundancy. But it's been what, 10 years? My dockstar shows no signs of giving up the ghost. Now that I see this issue isn't indicative of doom or anything. The exterior is the only part of it in bad shape! Its once glorious white finish is stained in a few places just due to mild exposure to UV light. Hilarious. I imagine at some point I might have to recap the brick power supply as I've done to a couple routers to bring them back.

Dockstar, a Kirkwood plug that won't die, mine has been running close to 10 years, keep on ticking like an Energizer bunny :))

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

bodhi Wrote:
-------------------------------------------------------
> Dockstar, a Kirkwood plug that won't die, mine has been running close to 10 years, keep on ticking like an Energizer bunny :))
>
+1 here. It is now used to host an Rpi0 (piggyback through its miniUSB port) and/or among other things.

I have an old and unused UG802 Mini PC -- a 2-core RK3066 with 1GB RAM. Perhaps, I should configure it to piggyback my Dockstar, too.