Software-Based Detecting and Recovering from ECC-Memory Faults

According to the problem that the ECC cannot correct the multibit error in ECC memory, this paper proposes a memory error processing method on software level. On the foundation of revising the Linux kernel code, the method can discover this area of influence area of memory error according to seek the process information mapping to the mistaken address. This way can avoid wastage to the user due to the system halting caused by memory error. The experimental results show that the method can have a certain degree of memory error repair and do not affect the normal work of the system.