DOC2000 erase size of 16k?
Quoted 4 lines
Er, that code is older than the 20010807 snapshot you said you were using. I meant the latest code from my CVS. I'm slightly concerned that the older code works when the newer code doesn't - it's supposed to work the other way round. Can you check the latest code and make sure it works there too?
Hmm, yes, the 20010812 snapshot works correctly. I just diff'd the two snapshots, and would have to say that the problems I experienced did not result from anything in the MTD code, so don't trouble yourself with what went wrong. I must have accidentally whacked something without noticing. My apologies for wasting your time.
Quoted 3 lines
It doesn't work perfectly yet, although it shouldn't take much to fix it. To make it fit, you have to use
nftl_formatto make the NFTL start further in to the device.
Ah! Now I see how this all fits together. Many thanks, and I will certainly write up my experiences for the HowTo. Incidentally, for anyone else trying this, it would serve them well to back up the bad blocks section of their DOC before doing *anything*. I seem to have killed my DOC by overwriting it, without a backup (fortunately, I can just get another from work, but not everyone has that luxury).
- Ben
Quoted 6 lines · [email protected] said:
Ah! Now I see how this all fits together. Many thanks, and I will
certainly write up my experiences for the HowTo. Incidentally, for anyone else trying this, it would serve them well to back up the bad blocks section of their DOC before doing *anything*. I seem to have killed my DOC by overwriting it, without a backup (fortunately, I can just get another from work, but not everyone has that luxury).
I don't believe it's possible to really kill the hardware. Just unwise to lose the factory-programmed information about bad sectors. We could fix nftl_format to preserve that information, and probably should do so quite soon.
What failure mode are you seeing with it now?
--
dwmw2