Cloudanix VulnDB

CVE-2026-90359

None

CVSS

—

EPSS

0.2%

Published

Sep 17, 2026

CISA due date

—

Cloudanix analysis

Schedule a fix SSVC: Track

A Linux kernel bug could let certain monitoring or security tools see or change only half the result when watching some functions, leading to wrong behavior or corrupted data.

Exploitation
Local Privilege
Attacker needs
Low Privilege
Complexity
Moderate
Impact
Data Tampering
Component
OS kernel
Class
Improper Input Validation
Remediation
Patch
Effort
Medium
Patch risk
Low

What it actually does

BPF trampolines preserve only 8 bytes of return values (RAX/R0) while accepting 16-byte __int128 returns for fexit/fmod_ret/fsession/struct_ops attach types; the second half (RDX/R3) is neither saved nor restored, causing corruption or partial observation of return values.

Why it matters to the business

Organizations using eBPF for networking, security enforcement, or observability may encounter unexpected data corruption or bypassed logic in production workloads when functions returning 128-bit values are instrumented, increasing operational risk until kernels are updated.

How an attack unfolds

An attacker compiles and loads a BPF program using an fexit, fmod_ret, fsession or struct_ops attach type on a kernel function that returns a 16-byte value; the trampoline mishandles the return, allowing the program to see or influence only half the value and corrupting the real caller’s view of the result.

How to fix it

Upgrade the Linux kernel to a version containing the fix that rejects >8-byte return values in bpf_check_attach_target(), bpf_check_attach_btf_id_multi(), and bpf_struct_ops_desc_init().

How to confirm the fix worked

After patching, attempt to load BPF programs with the affected attach types on functions returning 16-byte values and confirm the loads are rejected; verify the updated kernel source contains the new return-size checks.

Typical attack path

  1. Locate a target function returning __int128
  2. Load a BPF program via fexit/fmod_ret/fsession or struct_ops attach type
  3. Trigger the attached function to execute the trampoline path
  4. Observe or induce return-value corruption visible to both the BPF program and the original caller

How to detect it

Audit kernel logs and BPF program load events for attach types fexit, fmod_ret, fsession or struct_ops on functions known to return 128-bit values; look for subsequent anomalies in observed return values or application behavior.

What your security team should do

  • Prioritize kernel updates on all systems running eBPF workloads
  • Review existing fexit/fmod_ret/fsession/struct_ops programs for any that attach to functions returning large values
  • Add kernel version checks to CI pipelines that deploy BPF programs

The record

In the Linux kernel, the following vulnerability has been resolved: bpf: Reject >8 byte return values on return-reading trampoline paths btf_distill_func_proto() builds the function model used for the fentry/fexit/fmod_ret/fsession trampolines and struct_ops. It has accepted a 16-byte __int128 return value since the trampoline was introduced: __get_type_size() returns the integer's type size, and the return-type check only rejected ret < 0. But the BPF trampoline preserves only 8 bytes of the return value (RAX on x86, i.e. R0). For an attach type that reads the target's return value the second half (RDX / R3) is neither saved nor restored, so a program attached to a function returning a 16-byte value corrupts the value seen by the real caller and itself observes only half of it. struct_ops trampolines have the same limitation. This affects the attach types that read the target's return value: fexit, fmod_ret and fsession (plus the _multi variants of fexit and fsession), and struct_ops. fentry/fentry_multi run before the target returns and are unaffected. Reject a >8 byte return value for these attach types in bpf_check_attach_target() and bpf_check_attach_btf_id_multi(), and for struct_ops in bpf_struct_ops_desc_init().

Affected packages

Ecosystem Package Introduced Fixed