In the Linux kernel, the following vulnerability has been resolved: arm64: kprobes: Allow reentering kprobes while single-stepping A kprobe can be… (CVE-2026-98888)

The Linux kernel arm64 kprobes subsystem previously treated a kprobe hit in the KPROBE_HIT_SS (single-step) state as unrecoverable, similar to KPROBE_REENTER, which is the truly unrecoverable nested reentry case. This strict handling caused issues when a kprobe was hit while another was in the single-step state, such as during tracing or perf code execution from the debug exception path. The fix moves KPROBE_HIT_SS to the recoverable reentry cases, allowing the kernel to save the current kprobe state and set up single-step for the new probe. Additionally, the patch extends the prev_kprobe structure to save and restore the interrupt flag state (saved_irqflag) to prevent permanent interrupt disablement caused by overwriting the DAIF mask during nested kprobe handling. This change aligns arm64 behavior with a prior x86 fix.

Without this fix, nested kprobes during single-stepping could lead to incorrect handling of interrupt flags, potentially leaving interrupts permanently disabled on arm64 systems. This could cause system instability or degraded performance due to disabled interrupts. The vulnerability affects the kernel's internal debugging and tracing mechanisms but does not directly expose a remote attack vector or privilege escalation.

A fix has been implemented in the Linux kernel to correctly handle reentered kprobes during single-stepping on arm64 architectures and preserve interrupt state across nested kprobes. Users should update to the fixed kernel version once it is released. No additional mitigation is required beyond applying the official kernel patch when available.