Linux Kernel RAID5/6 Data Corruption Fix: Non-Accounting Mode Logic Error

Vulnerability Overview This vulnerability involves a logical error in the subsystem of the Linux kernel when handling RAID5/6 devices operating in non-accounting mode. When a RAID device is in non-accounting mode, the kernel fails to correctly distinguish between blocks that are "committed but not yet synchronized" and blocks that are "lost." This may cause the kernel to incorrectly mark data as lost during read failures, potentially leading to data corruption or system instability. Scope of Impact Affected Versions: Linux kernel versions 2.6.39 through 4.14 (specific versions must be confirmed based on the actual patch status). Affected Subsystem: The subsystem for RAID5/6. Affected Configuration: Only affects RAID devices configured in non-accounting mode. Remediation The fix introduces a new flag, , to distinguish between "committed but not yet synchronized" blocks and "lost" blocks. The specific modifications are as follows: 1. New Flag: Added the flag to the structure to identify whether a block has been committed but not yet synchronized. 2. Logic Adjustment: In the RAID5/6 read path, the block state is correctly determined based on the flag, preventing data from being erroneously marked as lost. 3. Code Modifications: - Modified the function in to include checks for the flag. - Modified the function in to ensure the flag is correctly set when submitting blocks. POC or Exploit Code The page does not contain specific POC or exploit code but provides the diff content of the fix patch, shown below: The code above illustrates the specific implementation details of the fix patch, ensuring that block states are correctly handled in non-accounting mode.

Goal Reached Thanks to every supporter — we hit 100%!

Linux Kernel RAID5/6 Data Corruption Fix: Non-Accounting Mode Logic Error