Impact
The kernel on LoongArch processors with the hardware page table walker enabled fails to propagate the _PAGE_DIRTY flag into the _PAGE_MODIFIED bit during write TLB misses. When a fork or clone operation triggers copy‑on‑write, the kernel’s pte_wrprotect() routine clears both the write and dirty bits, leaving the software dirty‑tracking bit unset. Consequently the page is considered clean during reclaim and may be freed without being written back, which can result in the loss or corruption of data that had been written to memory. The flaw represents a defect in the kernel’s handling of page‑table state and can lead to integrity failures.
Affected Systems
All versions of the Linux kernel that run on LoongArch CPUs with the hardware page table walker enabled and have not applied the patch referenced in the commit logs. No specific patch level is supplied, so any pre‑patch kernel that supports LoongArch and PTW is potentially vulnerable.
Risk and Exploitability
The CVSS score of 7.1 indicates medium‑high severity. The EPSS score of <1% suggests a low probability of exploitation. The vulnerability is not listed in CISA's KEV catalog, indicating no publicly known exploits to date. The likely attack vector is local, requiring a process to write to memory and then trigger a fork or a madvise(MADV_FREE) that creates page‑reclaim pressure. If successful, the kernel may free a dirty page without writeback, leading to data corruption.
OpenCVE Enrichment