Skip to contentSkip to conversation

"nand biterr" ... got an implementation?

U-Boot David Brownell 2 participants 3 messages Apr 22, 2009 – Jun 19, 2009

lore
Share & tools
lore
#0 lore raw

I was hoping that I could use that to test some NAND code, but then I noticed it wasn't implemented. I would have expected that U-Boot command wouldn't exist until they're implemented...

I'm hoping someone has an implementation. With at least these characteristics:

- Doesn't toggle the same bit each time
- Repeating N times on a block introduces N distinct 1-bit errors
- Normally affects only the data area, not spare/OOB
* although forcing bit errors in ECC would be valuable too
* ditto the rest of the OOB ... though likely they'd be ignored

Am I hoping in vain? ;)

- Dave

#1 lore raw
Quoted 3 lines · David Brownell

I was hoping that I could use that to test some NAND code, but then I noticed it wasn't implemented. I would have expected that U-Boot command wouldn't exist until they're implemented...

Sorry about that. :-(

It probably existed in an older version of the NAND code, and never got reimplemented.

Quoted 8 lines

I'm hoping someone has an implementation. With at least these characteristics:

- Doesn't toggle the same bit each time
- Repeating N times on a block introduces N distinct 1-bit errors
- Normally affects only the data area, not spare/OOB
* although forcing bit errors in ECC would be valuable too
* ditto the rest of the OOB ... though likely they'd be ignored

A general write-to-oob command would be nice as well.

-Scott

#2 lore raw

This is still in my mailbox, don't know that I'll get around to it any time soon, but another requirement came to mind.

Quoted 15 lines · Scott Wood
Quoted 3 lines · David Brownell

I was hoping that I could use that to test some NAND code, but then I noticed it wasn't implemented. I would have expected that U-Boot command wouldn't exist until they're implemented...

Sorry about that. :-(

It probably existed in an older version of the NAND code, and never got reimplemented.

Quoted 5 lines

I'm hoping someone has an implementation. With at least these characteristics:

- Doesn't toggle the same bit each time
- Repeating N times on a block introduces N distinct 1-bit errors

- Needs to reflect ECC structure. That is, when a 2K page is split into four distinct 512-byte ECC regions (plus OOB), it must be possible to force e.g. four errors in the second region while leaving the others alone.

Quoted 9 lines
Quoted 3 lines

- Normally affects only the data area, not spare/OOB
* although forcing bit errors in ECC would be valuable too
* ditto the rest of the OOB ... though likely they'd be ignored

A general write-to-oob command would be nice as well.

-Scott