---
id: "CVE-2026-98816"
kind: "vulnerability"
aliases: ["CVE-2026-98816"]
title: "Linux kernel BPF: constant blinding jump overflow crashes kernel with hardening"
severity: "unknown"
cvss: null
epss: null
exploited: false
status: "open"
category: "runtime"
products: []
ecosystems: []
tags: ["ebpf", "kernel", "linux", "runtime", "self-hosted"]
not_affected_if: ["bpf-disabled", "bpf_jit_harden-default"]
published_at: "2026-10-18T16:17:35+00:00"
first_seen_at: "2026-10-23T21:57:17+00:00"
updated_at: "2026-10-25T05:14:48+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-98816", "first_seen_at": "2026-10-23T21:57:17+00:00"}]
---

# Linux kernel BPF: constant blinding jump overflow crashes kernel with hardening

The Linux kernel's BPF JIT constant blinding rewrites a backward conditional jump and can overflow its signed 16-bit offset, so the jump lands past the end of the program. With bpf_jit_harden=2 the kernel crashes when such a program runs.

_Summary, check and fix written by a model from the sources; labelled as such._

## How to check

- Run `sysctl net.core.bpf_jit_harden` and check whether it is `2`
- Run `uname -r` and compare against the fixed kernel version listed by the distribution advisory (unknown from these sources)

## Fix

- Update the Linux kernel to a release that includes the upstream commit fixing bpf_jit_blind_insn() (version unknown from these sources)
- Keep `net.core.bpf_jit_harden` at `0` or `1` until patched

## Timeline

- 2026-10-23T21:57:17+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-98816

## From the sources

In the Linux kernel, the following vulnerability has been resolved:

bpf: Fix overflow of jump offset in constant blinding

bpf_jit_blind_insn() rewrites 'if rX op imm goto off' into:

  AX = imm ^ rnd
  AX ^= rnd
  if rX op AX goto off

and subtracts 2 from off when the jump is backward. off is s16, so
-32767 and -32768 wrap and the jump goes 32767 insns forward, past the
end of the prog. The jump is inside the patch, hence bpf_adj_branches()
doesn't check it. The kernel crashes when such prog runs with
bpf_jit_harden=2.

Return -ERANGE in such case. bpf_jit_blind_constants() fails the same
way when blinding of other insns moves a jump out of range.

---

_Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai_