A race condition exists between the dat_entry_walk() and large_crste_to_phys() functions in the KVM s390 implementation of the Linux kernel. If a large page is split concurrently, the code may dereference a stale page table entry and attempt to write a storage key to an invalid absolute address, leading to a crash of the kernel. The crash results in a system halt or reboot, effectively denying service for the affected node. No arbitrary code execution or data compromise is exposed by this flaw, but the impact on service availability is significant.

The vulnerability affects Linux kernel implementations running on the s390 architecture (IBM Z). No specific version range is provided, but the issue was fixed by recent kernel commits. Any systems that have not applied this patch and are running older kernel releases on s390 are potentially vulnerable.

No CVSS or EPSS score is publicly available for this vulnerability, and it is not listed in CISA KEV. The risk is therefore difficult to quantify precisely. The flaw requires a concurrent page split event that can be triggered by code executing in the context of the KVM. As the race is internal to the kernel, a local or privileged attacker would most directly exploit it; remote exploitation is unlikely. The primary consequence remains a denial of service via kernel crash rather than data breach or privilege escalation.