Hardware MicroSD data corrupts on the 3DS, but h2testw comes out fine

[Stellar]

This has been a bizarre week.

So, I have a 128gb Samsung microSD card. I've had it for years and it served me well through my previous 3DS, my Switch (until I upgraded to a 512GB card) and now I put it back into another 3DS I bought recently.

After I setup the 3DS with CFW I went and installed a bunch of games, including custom VC injects and DS roms which I load via Twilight Menu. Everything worked.

A few days ago I went to turn the 3DS on and it got stuck on a black screen with the blue LED on. A quick google search told me to run the "remove extdata" script and so I did. The 3DS booted and I thought nothing of it, until I realized some of my games wouldn't run anymore. They would crash on boot with ErrDisp messages.

I reinstalled the games and then they worked fine, but after I ran Twilight Menu I started noticing some of the ROMs had incorrect file sizes, and they too would not run. And when I tried to replace those ROMs, data would corrupt elsewhere.

My assumption was that the card became faulty, but after I ran a full 5 hour write+verify using h2testw, the card came back clean.

So now I have to ask what the hell could possibly be causing this?

 

[jamespoo]

sound like the 3ds is corrupting it so you might have to reinstall cfw or update it but my friend had a similar problem with a 64gb sd and told me he had to format the sd card to fat32 using http://www.sdcard.org/downloads/formatter_3/ and it solved the problem for him

 

[Stellar]

I formatted the card several times at this point, so I don't know what is causing the corruption to be honest.

I just did this, so fingers crossed.

But I wonder what caused this nightmare in the first place. I strongly suspect Twilight Menu.

 

[TurdPooCharger]

@Stellar Bull, when you used H2testw, did you full Write + Verify the SD card with it empty or blank?
You can only get a clean test run if the program completely fills up every space. Any preoccupied data can skew results by hiding defective sectors.

It might have been a good idea to use a data recovery software before reformatting and hardware testing. Data by now is completely wiped.

 

[ber71]

Cluster size? It was the same for me until I changed from 64k to the standard 32k cluster size. I believe twlmenu was behind that mess, too.

 

[Stellar]

@Stellar Bull, when you used H2testw, did you full Write + Verify the SD card with it empty or blank?
You can only get a clean test run if the program completely fills up every space. Any preoccupied data can skew results by hiding defective sectors.

It might have been a good idea to use a data recovery software before reformatting and hardware testing. Data by now is completely wiped.

The card was formatted prior to the test, so it was blank.

--------------------- MERGED ---------------------------

Cluster size? It was the same for me until I changed from 64k to the standard 32k cluster size. I believe twlmenu was behind that mess, too.

Cluster size has always been 32kb. Glad to hear TWL menu might in fact be the root cause. I'll just buy a flashcart.

 

[Stellar]

Well, after extensive testing it seems like everything works fine after the card format and I was even able to restore my backed up files and run each and every game.

The only thing I'm not doing is using TWLmenu again.

 

[TurdPooCharger]

Well, after extensive testing it seems like everything works fine after the card format and I was even able to restore my backed up files and run each and every game.

The only thing I'm not doing is using TWLmenu again.

Be on the look out for any further data corruption. H2testw only checks if the SD card's nand cells are intact.

Especially to SD cards that are older or heavily used from natural wear & tear, failures can be caused by dying controller. The SD card controller is like the teller in a bank who keeps track of how much money flows in and out between the customer and vault.

If the controller is going bad from heavy I/O, it would be like the teller finally cracking from too much work related stress.

Here's example of a faulty controller in NoSnake221's Kingston microSD, which did pass H2testw.

• N2DS SD card data management corruption? / start at [post #24] - read up to around [post #43].

 

[Stellar]

Be on the look out for any further data corruption. H2testw only checks if the SD card's nand cells are intact.

Especially to SD cards that are older or heavily used from natural wear & tear, failures can be caused by dying controller. The SD card controller is like the teller in a bank who keeps track of how much money flows in and out between the customer and vault.

If the controller is going bad from heavy I/O, it would be like the teller finally cracking from too much work related stress.

Here's example of a faulty controller in NoSnake221's Kingston microSD, which did pass H2testw.

• N2DS SD card data management corruption? / start at [post #24] - read up to around [post #43].

Thanks, though I noticed that fellow was also struck with data corruption whilst using Twilight Menu

 

[SparrowPT]

Man, I'm having this exact same issue and I'm pretty sure Twilight Menu is responsible.
My SD card also passes the H2testw with no errors. It's new and it always happens after using Twilight Menu for a while.
Just wanna raise some awareness as this is a serious issue and will stop me from using it ever again.

 

[delnirui]

It would be great if you filed a detailed report on this issue on TWiLight Menu++'s Github page so that this can hopefully be fixed.

 

[Deleted User]

(Deleted)