Bad Block Handling When formatting Nand flash

My understanding is the bad block information is stored in the sideband data, as well as the ECC to correct bad bits. So it's spread throughout the nand flash.

-Dave

As per my understanding the Bad Block information for Samsung is stored in the first and second page of every erase block.

During low-level formatting operation, this information must be preserved? How this was handled?

I don't think the chip erase wipes that information out. Is there a chip erase? It's been a while. If you erase sector by sector in your software, you've got to check the block/sector hasn't been marked bad from the factory before you touch it. If it was marked bad at the factory, leave it alone.

My knowledge of this comes from the u-boot project, which supports large nand flash devices with ECC + bad block marking. You could pull down the latest source there from git and look at the code, I think it's like cmd_nand.c.

I'm not sure but I don't think it deals with the situation of sectors going bad through normal use. The assumption is that with ECC even if a bit here or there goes bad, the data can still be read/written reliably. If there was a catastrophic failure of the memory, say 4 bits that just can't be written in a sector, it would break it (I think).

-Dave