NAND Flash: Bitflip and Read Only
Regarding the issue where NAND Flash bitflip and ECC errors cause blocks to become read-only, leading to excessive bitflips that cannot be corrected, it is speculated that this may be due to external interference.
During operation, if the device is in a harsh electromagnetic environment, data transmission from the SoC to the NAND is prone to bitflips, resulting in the NAND receiving incorrect or highly bitflipped data.
If multiple NAND Flash devices exhibit this issue, the following checks can be performed:
- Analyze the distribution of ECC fail addresses to see if they are concentrated in specific logical partitions. If concentrated in a specific logical partition, nearby areas, even if ECC passes, may show higher bitflip counts.
- Select one NAND Flash for testing and analysis, and scan the Vt of the error addresses. It is recommended to first confirm the corresponding addresses.
- Simultaneously compare the differences between "ECC data" and "correct data."
Bitflip
Possible causes include the following:
- Data is correct during writing, but bitflips occur during or after writing. Testing and analysis may reveal anomalies in the Vt of the erroneous data, and the errors on the NAND will appear more randomly distributed.
- Data is interfered with during writing, causing the NAND to receive incorrect data. Testing and analysis may show normal Vt, but the errors will be concentrated in specific logical partitions, such as frequently written partitions like data or logs. Scanning the entire logical partition may reveal many instances of high bitflip counts that still pass ECC.
Read Only
This issue may occur because the NAND Flash space has been filled without proper cleaning or due to long-term data storage that gradually fills up without periodic cleaning.
For spaces that are about to be filled, in scenarios where the space is already full or files have been stored for a long time without use, users need to confirm whether there is a relevant management mechanism to release or clean up space on the NAND Flash.
(GC Mechanism: In computer science, this refers to an automatic memory management mechanism. When a portion of memory occupied by a program is no longer accessed by that program, the program uses the GC algorithm to return that portion of memory to the operating system.)