HDD GURU FORUMS

Case:

I repair corrupt JPEGs, but these are corrupt because card is kaput (is my estimation). Basically bits flip all over the place, in header, throughout JPEG bitstream. So my guess is it's better to recover those files and honestly I don't feel like trying to figure out a pattern in corruption and write a custom tool for it.

Customer is wedding photographer in Dominican Republic. Who's close by and good with flash?

Suggestions appreciated.

ISTM that the problem is at the interface. Perhaps one of the DATx pins is unreliable. If so, would it be possible to revert to serial IO (SPI/MMC mode) rather than parallel?

https://pinouts.ru/Memory/sdcard_pinout.shtml

Carlos Marmolejos
https://www.datarecovery.com.do/
I have attended a training course with him.

@fzakbar: Thanks for suggestion but I don't have the card here. Just files.
@ddrecovery: Thanks! Mucha thanks!

Oh bummer, she already went there and they couldn't help.

Dammit.

So poking around a bit. Looking at JPEG header they're all corrupt in the same way, at same points. From recognizable text like Canon into Cankn and Exif into Exib and some more it looks as if for those bytes 6th bit flipped.

Jeremyb, coming your way

Arch Stanton wrote:From recognizable text like Canon into Cankn and Exif into Exib and some more it looks as if for those bytes 6th bit flipped.

Is it possibly an 8-bit card, eg MMCplus or MMCmobile?

have you tried reading with different voltage ?

Thanks. No. I haven't got the drive. Individual files were sent to me for repair. It is my estimation the files need to be properly recovered instead. I'm in the Netherlands, customer in Dominican Republic. So indeed need someone with proper equipment etc. to read the drive.

Arch Stanton wrote:Thanks. No. I haven't got the drive. Individual files were sent to me for repair. It is my estimation the files need to be properly recovered instead. I'm in the Netherlands, customer in Dominican Republic. So indeed need someone with proper equipment etc. to read the drive.

If Carlos cannot recover the drive I suspect it may not be recoverable.
Carlos has several of the professional DR Flash recovery softwares including PC-3000 Flash.
It sounds like the chip may have a lot of uncorrected ECC.

Thanks ddrecovery. Yes it sounds like that, but would that also explain that the observable corrupt bits appear to occur at the exact same places with the exact same bits flipped?

Again, I am not a hardware 'guy' and don't own any equipment to read flash other than a standard card reader. Just trying to gain some understanding.

Arch Stanton wrote:Thanks ddrecovery. Yes it sounds like that, but would that also explain that the observable corrupt bits appear to occur at the exact same places with the exact same bits flipped?

Again, I am not a hardware 'guy' and don't own any equipment to read flash other than a standard card reader. Just trying to gain some understanding.

Not sure what you mean by 'exactly the same places'.

Ah. Well these are photos shot using a Canon EOS. So I noticed in header Canon corrupts in to Cankn, EOS into EKS, Exif into Exib. In each of these files. You'd expect bit errors to be a bit more random, no?

Exif - Exib
01100110 - f changed into
01100010 - b

Canon - Cankn
01101111 - o changed into
01101011 - k

EOS - EKS
01001111 - O changed into
01001011 - K

Each time the 6th bit flips from 1 to 0.

Personally I am not aware of a specific hardware issue that would cause this. But a failing controller I guess could cause it, especially if the flash drive uses inversion. Have you tried using an image rebuilding program like disktuna.com.

lol.

Yes, that's how she ended up with me. I am disktuna.com. She asked if I can repair the files. But these flipping bits spread across the entire file. If I only flip back the ones in bitstream that result in invalid markers, after replacing header with a good one I get psychedelic stuff like this.

I just had a PM from a friend who is too shy to post on this forum......

He had a case where two data lines were shorted (D8 and D9) resulting in a very similar issue to this. I guess if it is every 6th bit which is flipped it would be possible to write a script to revert them.

Yes, got the PM too, thanks. Or another friend with the same idea, we will never know

But, it's not each and every 6th bit as far as I can tell. So I am still trying to see if there's some pattern. And if I find such a pattern I will have to flip back the entire card, I think it goes through the entire card.

But ideally I get an address where the lady can send the drive to. Because trying to get a disk image from her appears to be too much asked already. Someone who can d chip-off if needed maybe to bypass faulty electronics and such? That's beyond what I do, I don't have the equipment for that.

Arch Stanton wrote:But ideally I get an address where the lady can send the drive to. Because trying to get a disk image from her appears to be too much asked already. Someone who can d chip-off if needed maybe to bypass faulty electronics and such? That's beyond what I do, I don't have the equipment for that.

Carlos in the DR does have chip off capabilities, so sending it to someone else may not help.

Yes, so that is the dilemma, at least partially.

Give up assuming Carlos has tried everything? As I mentioned she has already been to Carlos, and her words: Carlos said the files were corrupt. Of course I wasn't there, I don't know what he actually said and tried. Maybe she told him I want to spend an x amount and Carlos decided it wasn't enough for going any further with this.

Since she still has the card and can use it I assume it hasn't been opened up etc..

There's also a bit of a language barrier going on as my native language isn't English and neither is hers.

BTW, I really appreciate your input and that of the others!