In the Linux kernel, the following vulnerability has been resolved: perf/core: Fill branch entries with a single assignment… (CVE-2026-98551)

The vulnerability in the Linux kernel perf subsystem was due to perf_clear_branch_entry_bitfields() clearing bitfields of struct perf_branch_entry individually but leaving some fields (new_type and priv) uninitialized. These fields were added later and not properly cleared, causing uninitialized memory data to be exposed in userspace on ARM64 systems when perf branch stack records were processed. The patch replaces per-field clearing with a single assignment of the entire struct, ensuring all fields are zeroed and consistent. The function perf_clear_branch_entry_bitfields() was removed as it had no callers left. This issue does not affect x86 architectures and is specific to ARM64 due to how the per-CPU branch stack is allocated and handled.

Uninitialized memory data from the perf branch stack could be exposed to userspace on ARM64 Linux systems, potentially leaking sensitive kernel memory contents. The vulnerability does not affect x86 architectures and is limited to the perf subsystem's branch stack handling. There is no indication of active exploitation in the wild.

A fix has been implemented in the Linux kernel that assigns the entire perf_branch_entry struct at once, ensuring all fields are properly initialized. Users should update to the kernel version containing this fix. Since this is a kernel-level issue, applying the official kernel update is the recommended remediation.