HDD GURU FORUMS

Can you please explain what exactly this condition is checked? I can't correct the image and I very often have a problem with this error, sometimes it is enough to slightly change the VCC or VCCQ voltage or the speed and the error stops appearing.

Out of curiosity, I checked the pages where this error is displayed and both the one read earlier (corrected) and the one that is being read are identical, apart from individual bit errors, so on what basis does the program decide that there is a "BCH reread error"?

In view of the above, I have an idea for the program to try to gently change the voltage +-0.1V, 0.2V before throwing this error and check if the chip responds correctly. If manipulating the voltages does not help, it could also try to change the speed (high, normal, low).

Additionally, the program could jump to the next sector/page/block and try to continue correcting the dump. Now I have to sit at the program and watch whether the error does not appear, I can't just leave the correction on overnight because at any moment this error can occur and interrupt the whole process - it is irritating.

Please consider this improvement - thanks.

btw. unchecking "Check chip response" is not a solution because unchecking it very often causes damage to the dump.

I have same question. Can somebody answer it? Thank you

Will someone from Rusolut answer my question and deal with this topic? I cannot efficiently process data recovery due to such a poorly functioning read retry.

If an error occurs, let the program simply jump to the next block. Instead of stopping the whole process. I noticed that these errors appear with damaged "garbage" blocks, if I skip these blocks by changing the start address, everything works OK. Just fix this by change the exception behavior from throwing an error to jumping to the next block. Thanks.

I've just ran into the exact same problem. Was about to run a Reread overnight but got BCH error immediately and at the same place everytime. I'm all in favor of a "if fail() then ignore()" instead of "if fail() then abort()".

Hello guys,
this error appears when a chip hangs and sends no data from a page where some data was read before. When a chip hangs it does not have any sense to proceed the reading, so the process stops, and you are supposed to check what is wrong, check the chip itself, check if it still alive at all, maybe try to find better parameters for reading. That is the idea.
However we will think about some additional further automations. Thank you for your input

Igor_Loskutov wrote:Hello guys,
this error appears when a chip hangs and sends no data from a page where some data was read before. When a chip hangs it does not have any sense to proceed the reading, so the process stops, and you are supposed to check what is wrong, check the chip itself, check if it still alive at all, maybe try to find better parameters for reading. That is the idea.
However we will think about some additional further automations. Thank you for your input

And this is a false assumption. The only logical explanation is that RR sets the value of one of the registers so that it causes an incorrect reading. And the answer "check the chip itself, check if it is still alive at all" irritates users since the chip works and reads normally (clean reading without RR parameters) so the problem is with RR - when I have "Reatr Retry not found" BCH error does not show up.

Read retry in VNR is one big mess, I don't know which RR to which chip, no control over the registers and their values. I would like to manually set the register values ​​and I can't, or ignore (not set) those that cause problems.

ps. VNR is "Visual" software, but when it comes to RR I deal with nothing visual, everything is hidden.

See how Flash Extractor had it done, you should also have RR done this way, where we can visually see the changing ECC colors (red, geen, gray) while searching for RR and so that we can see the values ​​of individual registers and edit them manually.

I think many will agree with me.

Gregory wrote:the problem is with RR - when I have "Reatr Retry not found" BCH error does not show up.

Sadly, BCH error also shows when using GPU without RR. Just had it with Sandisk BGA132.

melvin wrote:

Gregory wrote:the problem is with RR - when I have "Reatr Retry not found" BCH error does not show up.

Sadly, BCH error also shows when using GPU without RR. Just had it with Sandisk BGA132.

Speaking of ECC and GPU, it's not all rosy here either, you always need to make dump copies before running ECC. The ECC engine can destroy the entire dump and not detect that something is wrong. We had a case where the entire dump was overwritten with the values ​​0x5858 "XX" on RTX 4070Ti, but for example on RTX3060Ti everything was corrected fine.

This is quite old protection, that was introduced after several chips died during rereading and dumps were wiped. We did it to prevent such stories and it worked good so the assumption was correct. Now you report that it happens with alive chips (guys from our Support team confirmed it, but also said, that it happens really rare) and we will adjust the algorithm. Thank you for this

As for Read Retry, I think many will agree that they want to get as good as possible dumps with minimal involvement. So we did it this way, and it works good with most of the chips. If you have chips where read Retry Retry causes described problem, we will be glad to get them for testing.

Also, after some investigation we have found that the described problem may happen on cases where ECC is applied after XOR, with double logical page, on specific blocks. So could you please confirm that you get this problem on cases where ECC is applied BEFORE the XOR.

I right now have case where I get BCH reread error on ECC before XOR, ECC is on bBCR element but I don't think this is a problem. I have to jump to the next blocks/pages using start address to keep read going. I'll sign up for a ticket tomorrow and send it TV, you'll see.

ps. VNR set wrong RR for my chip, but even when a change to right RR have BCH error. And correct RR sometimes show RR found, sometimnes not found - is lotery. Maybe should be option for full scan register not only some, and base on this full scan doing adjustment for next pass.

BCH reread error, partial solved - chip must be reread using "Reread data blocks" to skip damaged blocks - .psf file must be available. I hope that in a future VNR update the skip damaged pages (white pages) option will be added - thats all. Thanks Igor for help. Regards.