Other Parts Discussed in Thread: AM3358
The whole machine cannot be turned on normally after abnormal power failure. After flashing the flash again, it works normally with a defect rate of 1% (approximately 10 units out of 1k pcs per month). It has been preliminarily confirmed that it is caused by damaged flash files.
The product software kernel adopts version 3.2.0 The client conducted the following tests and the related results are as follows:
1. When MLO reads uboot from nand, Hamming verifies that it can start normally;
2. Changing a certain bit in uboot, hoping that ECC verification could correct the error, but it did not, resulting in uboot being unable to start;
3. It seems that the ELM of AM335x is not working properly, resulting in ineffective NAND ECC during the uboot phase.
I hope to seek advice on the following questions:
1.What is the working mechanism of Elm? Is there any relevant documentation available to provide;
2. What are the common factors that cause ECC verification to be invalid;
3.I need to use the yaffs2 file system, but if I use BCH8, the oob area won't be enough. So I want to change it to BCH4. Why is BCH4 not supported? Only supporting BCH8, how to support BCH4, and what software changes should be made;
4.Will there be bit flipping when AM3358 reads NAND Flash?
5. How to implement a power-off protection software and hardware avoidance mechanism;
