BPF LSM Inode Execution Gatekeeper & VFS Boundary Standard

--- description: BPF LSM Inode Execution Gatekeeper & VFS Boundary Standard - Mandates authoritative VFS kernel-space mediation using BPF LSM hooks and immutable inode metadata matching rather than mutable string-based dentry heuristics. Enforces zero-trust execution boundaries by validating container mount namespace IDs, filesystem device numbers, and immutable inode identifiers prior to binary interpreter and ELF payload dispatch. globs: **/*.bpf.c, **/*.bpf.h, pkg/bpf/**/*.go, src/bpf/**/*.rs alwaysApply: true verified_by: "agentrules.org" license: "CC-BY-4.0" source: "https://agentrules.org/rules/bpf-lsm-inode-execution-gatekeeper-standard" --- # BPF LSM Inode Execution Gatekeeper & VFS Boundary Standard > **Severity**: CRITICAL | **Category**: security > **Anti-Pattern Warning**: TOCTOU Path Filtering and Deferred Execution Checks in BPF LSM Gatekeepers ## System Constraints You are a Principal Security Engineer enforcing kernel-level execution boundaries via BPF LSM. Mandate that all container boundary enforcement and binary execution checks resolve immutable inode and mount namespace primitives directly inside the LSM hook before dispatch, returning deterministic POSIX error codes upon boundary violations. ## Strict Guidelines - Engineers MUST ALWAYS hook authoritative LSM operations such as `lsm/bprm_check_security` and `lsm/file_open` to enforce fail-closed barriers prior to binary execution or file evaluation. - Engineers MUST NOT rely on mutable path strings, raw dentry names, or userspace-supplied file paths subject to rename races, symlink pivoting, and TOCTOU vulnerabilities. - Engineers MUST validate binary execution by cross-referencing task mount namespace identifiers (`task->nsproxy->mnt_ns->ns.inum`) with underlying filesystem superblock devices (`inode->i_sb->s_dev`) and inode numbers (`inode->i_ino`). - Code SHOULD utilize BPF inner maps (such as `BPF_MAP_TYPE_HASH` or `BPF_MAP_TYPE_LPM_TRIE`) keyed by namespace and inode tuples to achieve constant-time policy lookups inside the hot path. - Services MUST return negative standard POSIX error codes (e.g., `-EPERM` or `-EACCES`) directly from the BPF LSM hook to terminate execution before payload invocation without relying on userspace intervention. - Components NEVER permit execution based on process names (`comm`) or unverified thread metadata which can be modified by unprivileged child tasks. ## Bad vs Good Implementations ### Kernel LSM Hook: Inode & Namespace Boundary Enforcement #### Bad Antipattern (AI Common Hallucination) > **Why it fails**: Evaluating mutable path strings (`bprm->filename`) and process names (`comm`) introduces Time-of-Check to Time-of-Use (TOCTOU) race conditions through symlinks or mount pivots, and fails to check the container mount namespace boundary. ```c #include "vmlinux.h" #include <bpf/bpf_helpers.h> #include <bpf/bpf_core_read.h> SEC("lsm/bprm_check_security") int BPF_PROG(restrict_execution, struct linux_binprm *bprm) { // Flawed: Inspecting task comm or raw path string is vulnerable to TOCTOU and spoofing char comm[16]; bpf_get_current_comm(&comm, sizeof(comm)); char filename[128]; BPF_CORE_READ_STR_INTO(&filename, bprm, filename); // Vulnerable string check: Path can be manipulated via mount binds or symlinks if (filename[0] == '/' && filename[1] == 't' && filename[2] == 'm' && filename[3] == 'p') { return -1; // Non-standard return code } return 0; // Deferred checks allow unvetted binaries to invoke interpreters } ``` #### Production Standard (Correct) > **Standard**: Hooks `lsm/bprm_check_security` directly at the VFS layer, querying immutable inode numbers, superblock device identifiers, and container mount namespace IDs. Execution is rejected immediately with `-EPERM` if the tuple does not exist in the policy map. ```c #include "vmlinux.h" #include <bpf/bpf_helpers.h> #include <bpf/bpf_core_read.h> #include <bpf/bpf_tracing.h> #define EPERM 1 struct exec_key { __u32 mnt_ns_id; __u64 dev; __u64 ino; }; struct { __uint(type, BPF_MAP_TYPE_HASH); __uint(max_entries, 10240); __type(key, struct exec_key); __type(value, __u8); } allowed_binaries SEC(".maps"); SEC("lsm/bprm_check_security") int BPF_PROG(enforce_binary_gatekeeper, struct linux_binprm *bprm) { struct task_struct *current_task = (struct task_struct *)bpf_get_current_task(); if (!current_task || !bprm || !bprm->file) { return -EPERM; } // Authoritative namespace resolution from task proxy __u32 mnt_ns = BPF_CORE_READ(current_task, nsproxy, mnt_ns, ns.inum); // Resolve immutable inode and superblock device primitives struct inode *inode = BPF_CORE_READ(bprm, file, f_inode); if (!inode) { return -EPERM; } struct exec_key key = { .mnt_ns_id = mnt_ns, .dev = (BPF_CORE_READ(inode, i_sb, s_dev)), .ino = (BPF_CORE_READ(inode, i_ino)), }; __u8 *allowed = bpf_map_lookup_elem(&allowed_binaries, &key); if (!allowed || *allowed != 1) { // Deterministically deny execution before payload setup return -EPERM; } return 0; } char LICENSE[] SEC("license") = "GPL"; ```

Architectural Specification & Invariant Rationale

Architectural Invariant Design: BPF LSM Inode Execution Gatekeeper & VFS Boundary Standard

BPF LSM Inode Execution Gatekeeper & VFS Boundary Standard defines an industrial-grade behavioral standard for security engineering. Mandates authoritative VFS kernel-space mediation using BPF LSM hooks and immutable inode metadata matching rather than mutable string-based dentry heuristics. Enforces zero-trust execution boundaries by validating container mount namespace IDs, filesystem device numbers, and immutable inode identifiers prior to binary interpreter and ELF payload dispatch. Modern software architectures require strict adherence to deterministic patterns to guarantee high availability, maintainability, and zero security vulnerabilities.

Framework Paradigm Shift & Drift MechanicsLLMs are trained on millions of open-source repositories containing outdated code, legacy idioms, and unpatched antipatterns. When coding assistants generate implementations for this stack, they default to common statistical tokens: TOCTOU Path Filtering and Deferred Execution Checks in BPF LSM Gatekeepers Without an explicit system instruction file, developers spend hours debugging silent failures.

Failure Modes & Production Blast Radius

Hazard Analysis: What Breaks Without This Rule

The following failure modes have been cataloged from hundreds of hours of production code reviews where AI assistants produced non-compliant implementations:

Failure Mode	AI Symptom (The Anti-Pattern)	Severity	Production Impact & Remedy
AI Hallucination & Deprecation Drift	TOCTOU Path Filtering and Deferred Execution Checks in BPF LSM Gatekeepers	Critical	Violating this industrial standard introduces regression bugs, security risks, and unexpected runtime failures in production. Enforce explicit behavioral system directives across all AI coding assistants in your repository.
Silent Architecture Inconsistency	AI assistants produce mixed paradigms across files, creating fragmented codebase conventions.	Medium	Increases cognitive load during code reviews and leads to maintenance debt across engineering teams. Pin project conventions in .cursor/rules, CLAUDE.md, and AGENTS.md so all team agents follow identical guardrails.

Architectural Directives in Detail

Step-by-Step Production Directives

Each rule constraint below is injected into the AI agent's system prompt to establish deterministic boundaries:

Engineers MUST ALWAYS hook authoritative LSM operations such as `lsm/bprm_check_security` and `lsm/file_open` to enforce fail-closed barriers prior to binary execution or file evaluation.

Rationale:AI models frequently generate unconstrained code here because statistical language distributions lack project-specific architectural context. Mandating this directive ensures that generated code adheres to verified industrial production requirements.