Fix SMART attribute UDMA_CRC_Error_Count on a WD Red

[Oungawak]

Hello,

I have a very annoying problem and I hope I came to the right place.

I have a WD Red HDD that previously was in a faulty external case. I moved it to a Synology NAS in RAID 1 with another HDD of the same model and since then the NAS worked fine. But the NAS do regular SMART checks and keep sending me warnings about that drive, more specifically it complains the disk had 16 'reconnections' because the SMART attribute "UDMA_CRC_Error_Count" says so.

I checked what this attribute means and apparently this error can come, NOT from the HDD itself, but from a faulty connection, and this can be confirmed by replacing the connection and checking if the error count stops increasing. I checked and I can confirm this value completely stopped increasing after I moved the HDD to the NAS.

So the HDD seems OK but the SMART data on it have been 'tainted' by the faulty external case, and now SMART checks incorrectly fails on the NAS. Not a very good check IMO...

It seems there is no way to ignore/hide this error on the Synology NAS, so now I just want to reset this useless error count. My searchs found it's not trivial but it can be done with the right software and the appropriate "Vendor Specific Command". But so far I haven't found how to do that yet for my HDD.

Could you help me reset this "UDMA_CRC_Error_Count" attribute on my WD Red please? What software do I need and what steps should I follow?

Thank you very much.

[fzabkar]

Are you able to read all the firmware modules with WDMarvel (demo) or with a HDDSuperTool script?

[Oungawak]

Hello fzabkar and thank you for helping me!

In my previous search I found WDMarvel and I managed to launch the demo on a 32-bits Windows 10 I installed on a old machine (someone said those kind of tools don't work well on 64-bits Windows). But I will be honest, I still can't make any sense of what this software shows me... So could you point me where to find those firmware modules?



This is why I prefer asking for help, those tools looks very specialized and touch the firmware, so I don't want to brick my disk over a silly mistake. About that, for now I do my tests on an old WD Green in bad health but still working (I don't care if I break this one). Do you think I can keep experimenting on it and apply the same method on the WD Red? It would be safer and simpler for me, as I don't want to pull the WD Red from the NAS until I know what to do with it.

For HDDSuperTool I haven't tried it yet as it requires Linux, but if you think it's a better tool I can boot a LiveCD or install Debian on the old machine. I don't have any script for it though.

[fzabkar]

You should find the option to read the SA modules under "S/A Operations".

[diybit]

Try what suggest fzabkar, and done a full backup of your current modules by reading them, also done a checksum check of all modules to be sure of the health state.

- Disable your SATA channels/controllers inside the "Windows Device Manager" (this will avoid OS from doing some "bullshit" on the port, that can alter the data you read/write)
- Identify the drive by choosing the right port and clicking "Detect"
- Read modules : S/A operation -> Modules -> read all modules, and then check all
- Also backup your rom : ROM -> Read

Then, I suggest you trying the vendor specific command Format Unit, that will normally done a tiny lifting of your drive and resetting some SMART attributes, the CRC error count will normally be reseted. (but I dont know if this format will work on this WD drive)

There is a tool that seems doing that : HGST_fmt.exe that you can found inside this thread http://www.hddoracle.com/viewtopic.php?f=21&t=2068
I have never tested this tool, so wait someone to confirm you using it, or maybe to get another alternative way... (maybe without formatting...)

Here another link that can interest you https://forum.hddguru.com/viewtopic.php?f=7&t=35736

What is your drive model name?

Last edited by diybit on November 12th, 2020, 16:09, edited 1 time in total.

[legija]

Set SATA controller to AHCI mode in Bios.
On WDMarvel select desired drive and tick "lock current device"
Menu -> Buttons,
Modules -> Restore Points
Create point.

[Oungawak]

Hello again everyone,

For the model names:
- WD Red I want to fix -> WDC WD60EFRX-68MYMN1
- WD Green I use for tests first -> WDC WD15EADS-00P8B0

I'm not confident enough yet to fiddle with my WD Red, but I made a lot of progress learning how to use WDMarvel with my test HDD (the WD Green in bad health that I don't care about).

Here is what I did so far with the test HDD:
- Isolate the HDD from Windows -> Disable the HDD and all SATA controllers in Device Manager (one controller can't be disabled, probably used by the system drive?), Disk Manager don't see the HDD anymore so I think we are safe
- Select the right HDD in WDMarvel -> In the "IDE/SATA controller selection" section, select in the menu on the right the SATA mode used by the motherboard ("SATA/AHCI" in my case) and select a SATA port in the grid, then in the "Drive" section, click the "Detect" button to check if the port is for the right drive (Windows gets neauseous if you stumble upon the system drive, I rebooted just in case...)
- Once the right port is found, check "Lock current device" (not sure why, I suppose it's so WDMarvel gets some sort of exclusive access to the HDD I guess?)
- Backup the firmware

For the backup I used diybit's method:
- Click on the menu "S/A operations > Modules operations"
- In the new module window click on "Modules > Read all" (modules are dumped in the folder "Default\Modules" in WDMarvel directory)
- Once the operation is finished click on "Modules > Check all modules"
- Close the module window
- Click on the menu "ROM > Read ROM" (ROM is dumped in file "Default\PZU\Flash.bin" in WDMarvel directory)

The health of the WD Green seems bad as expected, because the module 6F ("Overlay") have an invalid CRC. But I use it only to test WDMarvel anyway so I don't care.

After that I used legija's method, I clicked on the menu "S/A operations > Restore points" and on the button "Create point". Apparently it does the same dumps, but with more data and neatly packaged in a ZIP archive in "Default\RestorePoints".

So far I think I finished all the preparations, but I'm not sure what to do next. I found in the menu a "Format unit" like diybit suggested if someone could confirm this is the right option, and I also found 2 "clear SMART" features that looks rather explicit:
- Defects > PDList and format > Format unit (Cmd)
- Heads > Clear SMART, defect lists, Translator
- Other > S.M.A.R.T. > Clear S.M.A.R.T.

Could you tell me which one should I choose?

Also the WD Green as been a good test subject until now but it doesn't have the SMART error I try to fix on the WD Red. Do you have a way to check if SMART data have been reset? If not well at least it will confirm what I do in WDMarvel don't kill the HDD right away.

[diybit]

Your config seems right, all channels disabled, and me too I let one controller or I cant communicate with the drive, and yes the drive disappear from Windows, but can be handle by WDMarvel or others.

It is normal that module 6F got invalid CRC. (and about health, it is just about SA health Im talking about)

It is not sure that what you try on the Green will give identical result as trying on the Red. (especially for the VSC "Format Unit")

The VSC "Format Unit" Im talking about is not include inside WDMarvel. (as I know...)
It is normally a VSC for Hitachi/HGST/IBM drives, but I have read somewhere that it can works on "new" WD drive also, so your drive will maybe understand it.
To try HGST_fmt.exe : finally the file is HGST_eng.exe from Hitachi_format.rar and combine it with Io.dll from Io.rar, these files are in the hddoracle thread.

My suggestion is for in case you dont have a full version of WDMarvel, or for in case you dont want to done a complete SMART reset.
If you get the full version, you can simply done "Other > Clear SMART" and that's all, but all attributes will be reset.

Im very curious on what will be the solution proposed by fzabkar... sorry to have maybe speak a little too speedy and little cut you on what you would done fzabkar

[fzabkar]

I would edit the SMART module(s) and write them back to the SA. Alternatively there is a way to hide a particular SMART attribute, but I don't know if hidden attributes would be ignored by the SMART check.

I have written a tool to display the SMART attributes. It can also "unhide" hidden attributes. I believe WDMarvel can do this now, too.

http://www.users.on.net/~fzabkar/FreeBasic_W32/WD/wdsmart.bas
http://www.users.on.net/~fzabkar/FreeBasic_W32/WD/wdsmart.exe

[diybit]

Haha can be funny to solve it by hiding the CRC error count. (but maybe not recommended to spot future CRC error)
Do you have found a solution to do this hiding without having to full clear SMART?

About rewriting the SMART modules, from some experiments I have done with your help fzabkar, it was impossible to customize the raw values of the SMART.
But maybe I have not well disabled the SMART before trying it, recently I have think to another way to disable it that I have to try one day...
I also read that this simple module rewriting can also works on some drive...

So Oungawak, if you clear SMART, it's likely you cant go back.
I add that I finally dont know if those format you see inside WDMarvel will clear the CRC error count, not tested with this CRC error situation, you can eventually give a try for we know.
Do you have WDMarvel full version?

[Oungawak]

Unfortunately I only have the demo version of WDMarvel, I'm just doing this with nothing but an old PC laying around. I forgot that, maybe the features I pointed out are locked. The website says the demo is "read-only" so WDMarvel could be a dead end.

About resetting SMART: It's an important feature and I don't want to break it too much. If possible I would prefer a "surgical strike" approach instead of a full reset, I don't want the disk to declare itself "brand new" erroneously. On the other hand about "UDMA_CRC_Error_Count", it would be safer to only reset it, but if we have to kill/hide it... Honestly it may as well be a good idea, it seems to do more harm than good anyway... I still wonder why this counter even exists in the first place considering how much troubles it causes, and I've seen on the web I'm not alone with that problem.

I got diybit's format program, but it asks for a "Command port" that I don't know (with a "Slave" checkbox, for IDE ports maybe?). I tried entering "0", "1", "2", etc. but the "Identify HDD" button doesn't output any meaningful data. Either I need to find the right value for "Command port", or this tool simply don't like the WD Green... This is an older drive than the WD Red though, so if this program works with "new" WD only maybe it will work on the WD Red?

Also could you tell me more about what kind of "tiny lifting" this program do?

[fzabkar]

ISTM that the problem with modifying SMART modules is that there is a running copy in RAM, and this is periodically written back to the SA. So after you modify the SA, the drive undoes your changes with the original SMART data in RAM.

[diybit]

And about your patching method to unhide/hide attributes, did you found another way that do not need clearing SMART to apply it? (with a modified mod23)

[fzabkar]

Just manually edit the raw data in the appropriate SMART module.

[diybit]

So and we re fall to the problem while rewriting SMART module...
How do you success you to bypass this RAM copy problem?

[fzabkar]

If you can't stop the firmware from updating the SMART module from its RAM copy (I'm only guessing that this is what is happening), another way which we tried at the HDD Oracle was to ...

find an unused area of the SA
write the SMART modules to this area
edit module 01 (DIR) to point to this unused area
power cycle the drive
write the SMART modules to the original area in SA
restore original module 01
power cycle the drive

There are two checksums that need to be recalculated in each SMART module.

[diybit]

Okay fzabkar! I well see now where we go with this solution, very thanks to have written it!!
And sorry to have little re-ask you what I already do year ago, that finally lead to same solution.
I would to be sure, and previously I was stuck on this clear SMART method with mod23 template, so it's for that that I was not understanding why moving the SMART module would change something.
But it's totally different if we just modify/move the main SMART module.
Humm you say we need to also modify others SMART modules, it is just to reflect some change accordingly to what we made in the main? (like unhide/hide)


So Oungawak, without a full WDMarvel, cross your fingers that this VSC "Format Unit" will works... (but there may be others ways with the help of HDDSuperTool or MHDD scripts...)
The "Command port" you have to enter is the one used inside WDMarvel to detect your drive, col "Cmd port" like D030.
Previously to start using HGST_eng.exe, unplug power of your drive, and replug it, directly under Windows. (and avoid same plug cable for your two drives)
If it do not detect at your entered port, it can also be because of your AHCI mode, try pass to IDE mode.
I re-said, I have never used this tool, it is surely safe but be warn, dont try it with important data on primary drive.
It's not sure if the tool will right indicate if success/fail doing the VSC, so check your HDD led, and touch/hear your drive to see if it is doing something, it's a long format normally, about 3-4hours on your Green.
If your Green success, the Red will about also. But maybe the Green will not, and the Red yes...
The only WD drive I have used with this VSC is an "old" SATA Raptor WD740, and it have succeeded resetting CRC error, but it's an very strange WD drive compared to yours.

The tiny lifting is a true low level format that will move the defects sectors found or already marked to the hidden primary list (P-List).

[diybit]

Do a screen/save of your current SMART values before starting the format to compare after.
You said the Green is defective, does it show pending sectors or reallocated inside his SMART?

[BGman]

ID20=SMART & Log Current
ID21=SMART & Log Current
ID22=SMART & Log Current
ID23=SMART & Log Clear
ID24=SMART & Log Main

ID2D=SMART Log - Dump RAM
ID2E=SMART Log - Dump RAM

[Oungawak]

Don't worry about my data, the WD Green is empty and the WD Red is in a RAID1 on my NAS, with a backup of the NAS elsewhere, of course. The only thing I fear is bricking the WD Red, I would survive the loss financially but I would be disappointed.

I don't have a "Cmd port" column so I tried what is in the "Port address" column (DFE3A180), but all I get is "<HDD> Error: HDD not identify". Is it still the wrong ID or does it means the WD Green is incompatible?

Here are the WD Green infos in WDMarvel:


I can't switch from AHCI to IDE mode either. Windows boots normally then shows a blue screen with "INACESSIBLE_BOOT_DEVICE".

Also if the HDD needs a low level format, maybe I could use others formating tools meant for WD? I found this one for example that seems to be free, compatible with WD and do low level formats:
https://hddguru.com/software/HDD-LLF-Lo ... rmat-Tool/

About the health of the WD Green, the raw SMART value for "Pending Sector" is 1465, and for "Uncorrectable Sector" it is 1387, which seems high for me, but this is not why I stopped using it. If I remember correctly (it was long ago, and I was WAY less experienced, especially on backup practices ) I got problems with my data on it: Corrupted files and I think Windows froze when accessing some folders... Back then I somehow managed to mitigate the problem by creating a dummy 2GB partition at the start of the disk before the real partition, but of course I took the safe option and just replaced it when I could. I kept it around though as it's not completely dead yet. A good idea considering how it helped me in my experiments. It would be funny if a low level format fixed it, but I don't think I could trust it again.

Here are the SMART data from WDMarvel:


And here is the G-List:


The P-List is absolutely massive with 4904 defects, but from what I learned those are defects from the factory and can be ignored.