The Linux kernel contained a bug in the AMDGPU driver where the minimum size for UVD decode images was calculated using the pitch value instead of the width. This logic error could allow a pitch value exceeding 4096 to cause an integer overflow, potentially leading to a buffer overflow or denial of service. The commit that fixed the issue also adds a rejection for pitches larger than 4096 to prevent the overflow. The vulnerability was addressed before release, and the provided references show the patch implementation.

All Linux distributions that ship the Linux kernel before the commit that fixed the UVD decode image size calculation are affected. The issue manifests only in environments that use the AMDGPU driver and attempt to decode UVD streams. If a system uses a kernel version that incorporates the commit, the bug is resolved; otherwise any older kernel is vulnerable.

No CVSS score is reported, and the EPSS is not available, so the risk is unquantified but the flaw is a classic integer overflow leading to a potential memory corruption. The vulnerability was not listed in CISA KEV, indicating no known active exploitation. Based on the description, it is inferred that an attacker could trigger the overflow by sending a video decode request with a pitch greater than 4096, likely requiring local or privileged execution. The impact would be a crash or potential escalated privileges depending on the context of the memory write.