mirror of
https://github.com/elicpeter/nyx.git
synced 2026-06-24 20:28:06 +02:00
[pitboss] phase 25: Track G.2 — Path search, scoring, ChainFinding emission, SARIF property
This commit is contained in:
parent
a3ab1215f1
commit
76d0037073
12 changed files with 1908 additions and 139 deletions
202
src/chain/finding.rs
Normal file
202
src/chain/finding.rs
Normal file
|
|
@ -0,0 +1,202 @@
|
|||
//! Phase 25 — chain finding emitted by the composer.
|
||||
//!
|
||||
//! A [`ChainFinding`] is the externally-visible artefact produced by
|
||||
//! Track G: a sequence of static findings whose composition implies a
|
||||
//! higher-level [`ImpactCategory`] than any single member. The chain
|
||||
//! has its own [`ChainSeverity`] (a strict superset of the per-finding
|
||||
//! [`crate::patterns::Severity`] axis, with `Critical` reserved for
|
||||
//! chains so default-severity gates do not accidentally fire on a
|
||||
//! chained-only impact).
|
||||
//!
|
||||
//! # Determinism
|
||||
//!
|
||||
//! `stable_hash` is the BLAKE3-truncated digest of the chain member
|
||||
//! hashes joined with the implied impact byte. Two scans of the same
|
||||
//! source produce the same `stable_hash` regardless of DFS visitation
|
||||
//! order.
|
||||
//!
|
||||
//! # Suppressing constituents in default output
|
||||
//!
|
||||
//! Phase 25 keeps individual constituent findings on the wire — they
|
||||
//! still travel inside `Diag` form — but the JSON / SARIF emitters
|
||||
//! gate their visibility on [`crate::utils::config::OutputConfig::show_chain_constituents`].
|
||||
//! See `crate::output::filter_constituents` for the gating.
|
||||
|
||||
use crate::chain::edges::FindingRef;
|
||||
use crate::chain::impact::ImpactCategory;
|
||||
use crate::evidence::VerifyResult;
|
||||
use serde::{Deserialize, Serialize};
|
||||
use std::fmt;
|
||||
|
||||
/// Severity bucket assigned to a [`ChainFinding`].
|
||||
///
|
||||
/// Distinct from [`crate::patterns::Severity`] so that chain output
|
||||
/// (which is, by construction, a composition of *several* findings)
|
||||
/// does not collide with the per-finding axis. `Critical` is the
|
||||
/// highest grade and is reserved for chains whose impact is
|
||||
/// terminal RCE (`Rce`, `BrowserToLocalRce`).
|
||||
#[derive(Debug, Copy, Clone, Eq, PartialEq, Ord, PartialOrd, Hash, Serialize, Deserialize)]
|
||||
#[serde(rename_all = "snake_case")]
|
||||
pub enum ChainSeverity {
|
||||
Low,
|
||||
Medium,
|
||||
High,
|
||||
Critical,
|
||||
}
|
||||
|
||||
impl fmt::Display for ChainSeverity {
|
||||
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
|
||||
f.write_str(match self {
|
||||
ChainSeverity::Low => "LOW",
|
||||
ChainSeverity::Medium => "MEDIUM",
|
||||
ChainSeverity::High => "HIGH",
|
||||
ChainSeverity::Critical => "CRITICAL",
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
/// One member of a [`ChainFinding`].
|
||||
///
|
||||
/// Wraps a [`FindingRef`] so the chain output can name each constituent
|
||||
/// without duplicating the finding's evidence; consumers join back to
|
||||
/// the `findings: [...]` array via [`FindingRef::finding_id`] /
|
||||
/// [`FindingRef::stable_hash`].
|
||||
pub type ChainMember = FindingRef;
|
||||
|
||||
/// A composed exploit chain.
|
||||
///
|
||||
/// Phase 25 emits these from [`crate::chain::search::find_chains`].
|
||||
/// Phase 26 will populate `dynamic_verdict` from a composite
|
||||
/// re-verification pass; Phase 25 always leaves it as `None`.
|
||||
///
|
||||
/// `PartialEq` is omitted because [`crate::evidence::VerifyResult`] is
|
||||
/// not `PartialEq`. Equality checks at the test layer compare on
|
||||
/// `stable_hash` instead.
|
||||
#[derive(Debug, Clone, Serialize, Deserialize)]
|
||||
pub struct ChainFinding {
|
||||
/// BLAKE3 of `(member.stable_hash for member in members) || implied_impact`,
|
||||
/// truncated to 64 bits. Stable across scans for the same chain.
|
||||
pub stable_hash: u64,
|
||||
/// Constituent findings, in path order (entry-adjacent first,
|
||||
/// sink-adjacent last).
|
||||
pub members: Vec<ChainMember>,
|
||||
/// The dangerous-local sink terminating the chain. Carries the
|
||||
/// callee function name and cap bits so consumers can describe
|
||||
/// the chain without re-walking the SurfaceMap.
|
||||
pub sink: ChainSink,
|
||||
/// Composed impact category derived from member caps + adjacency.
|
||||
pub implied_impact: ImpactCategory,
|
||||
/// Chain severity, computed in [`crate::output::severity`].
|
||||
pub severity: ChainSeverity,
|
||||
/// Numeric score from [`crate::chain::score::score_path`].
|
||||
/// Carried verbatim for JSON output so consumers can re-sort.
|
||||
pub score: f64,
|
||||
/// Composite dynamic verification verdict. `None` in Phase 25
|
||||
/// (the composite re-verifier lands in Phase 26).
|
||||
#[serde(default, skip_serializing_if = "Option::is_none")]
|
||||
pub dynamic_verdict: Option<VerifyResult>,
|
||||
}
|
||||
|
||||
/// Sink terminus of a [`ChainFinding`]. Mirrors the
|
||||
/// [`crate::surface::DangerousLocal`] node the path ends at.
|
||||
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
|
||||
pub struct ChainSink {
|
||||
pub file: String,
|
||||
pub line: u32,
|
||||
pub col: u32,
|
||||
pub function_name: String,
|
||||
pub cap_bits: u32,
|
||||
}
|
||||
|
||||
impl ChainFinding {
|
||||
/// Compute the stable hash from a member list + impact category.
|
||||
/// Exposed so callers that build a `ChainFinding` outside
|
||||
/// [`crate::chain::search`] (tests, future composers) stay in sync
|
||||
/// with the canonical hash formula.
|
||||
pub fn compute_stable_hash(members: &[ChainMember], implied_impact: ImpactCategory) -> u64 {
|
||||
let mut h = blake3::Hasher::new();
|
||||
for m in members {
|
||||
h.update(&m.stable_hash.to_le_bytes());
|
||||
}
|
||||
h.update(&[impact_byte(implied_impact)]);
|
||||
let out = h.finalize();
|
||||
let bytes = out.as_bytes();
|
||||
u64::from_le_bytes(bytes[..8].try_into().unwrap())
|
||||
}
|
||||
}
|
||||
|
||||
/// Stable byte tag for each [`ImpactCategory`]. Used by
|
||||
/// [`ChainFinding::compute_stable_hash`] so adding an impact variant
|
||||
/// does not silently shift every other chain's hash.
|
||||
const fn impact_byte(c: ImpactCategory) -> u8 {
|
||||
match c {
|
||||
ImpactCategory::Rce => 1,
|
||||
ImpactCategory::BrowserToLocalRce => 2,
|
||||
ImpactCategory::SessionHijack => 3,
|
||||
ImpactCategory::InternalNetworkAccess => 4,
|
||||
ImpactCategory::InfoDisclosure => 5,
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use crate::chain::edges::FindingRef;
|
||||
use crate::surface::SourceLocation;
|
||||
|
||||
fn member(hash: u64) -> ChainMember {
|
||||
FindingRef {
|
||||
finding_id: format!("f-{hash}"),
|
||||
stable_hash: hash,
|
||||
location: SourceLocation::new("a.py", 1, 1),
|
||||
rule_id: "test".into(),
|
||||
cap_bits: 0,
|
||||
}
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn stable_hash_changes_with_member_order() {
|
||||
let a = ChainFinding::compute_stable_hash(
|
||||
&[member(1), member(2)],
|
||||
ImpactCategory::Rce,
|
||||
);
|
||||
let b = ChainFinding::compute_stable_hash(
|
||||
&[member(2), member(1)],
|
||||
ImpactCategory::Rce,
|
||||
);
|
||||
assert_ne!(a, b);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn stable_hash_changes_with_impact() {
|
||||
let a = ChainFinding::compute_stable_hash(
|
||||
&[member(1), member(2)],
|
||||
ImpactCategory::Rce,
|
||||
);
|
||||
let b = ChainFinding::compute_stable_hash(
|
||||
&[member(1), member(2)],
|
||||
ImpactCategory::BrowserToLocalRce,
|
||||
);
|
||||
assert_ne!(a, b);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn stable_hash_deterministic_across_calls() {
|
||||
let h1 = ChainFinding::compute_stable_hash(
|
||||
&[member(1), member(2), member(3)],
|
||||
ImpactCategory::Rce,
|
||||
);
|
||||
let h2 = ChainFinding::compute_stable_hash(
|
||||
&[member(1), member(2), member(3)],
|
||||
ImpactCategory::Rce,
|
||||
);
|
||||
assert_eq!(h1, h2);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn severity_ordering_is_critical_top() {
|
||||
assert!(ChainSeverity::Critical > ChainSeverity::High);
|
||||
assert!(ChainSeverity::High > ChainSeverity::Medium);
|
||||
assert!(ChainSeverity::Medium > ChainSeverity::Low);
|
||||
}
|
||||
}
|
||||
|
|
@ -34,11 +34,17 @@ use serde::{Deserialize, Serialize};
|
|||
|
||||
pub mod edges;
|
||||
pub mod feasibility;
|
||||
pub mod finding;
|
||||
pub mod impact;
|
||||
pub mod score;
|
||||
pub mod search;
|
||||
|
||||
pub use edges::{ChainEdge, FindingRef, findings_to_edges};
|
||||
pub use feasibility::Feasibility;
|
||||
pub use finding::{ChainFinding, ChainMember, ChainSeverity, ChainSink};
|
||||
pub use impact::{IMPACT_LATTICE, ImpactCategory, ImpactRule, lookup_impact};
|
||||
pub use score::{ChainScoreConfig, category_weight, min_score_default, score_path};
|
||||
pub use search::{ChainSearchConfig, find_chains};
|
||||
|
||||
/// One node in a [`ChainGraph`].
|
||||
///
|
||||
|
|
|
|||
192
src/chain/score.rs
Normal file
192
src/chain/score.rs
Normal file
|
|
@ -0,0 +1,192 @@
|
|||
//! Phase 25 — scoring for composed exploit chains.
|
||||
//!
|
||||
//! `score(path) = sum(impact) * product(feasibility)`
|
||||
//!
|
||||
//! The impact term is the sum of per-member [`ImpactCategory`] weights
|
||||
//! (each member contributes the weight of the *standalone* category its
|
||||
//! primary cap maps to, or `0` when the cap has no standalone impact —
|
||||
//! the cap still contributes adjacency to the final implied impact via
|
||||
//! the composer). The feasibility term is the product of every
|
||||
//! member's [`Feasibility::score`].
|
||||
//!
|
||||
//! # Threshold
|
||||
//!
|
||||
//! [`min_score_default`] is the in-code fallback when `[chain] min_score`
|
||||
//! is unset in `nyx.toml`. Path search drops any composed chain whose
|
||||
//! score is strictly below the configured threshold.
|
||||
|
||||
use crate::chain::edges::ChainEdge;
|
||||
use crate::chain::feasibility::Feasibility;
|
||||
use crate::chain::impact::ImpactCategory;
|
||||
use serde::{Deserialize, Serialize};
|
||||
|
||||
/// Per-impact-category numeric weight contributed to the additive
|
||||
/// impact term. The relative ordering matches the design doc's
|
||||
/// criticality ranking; absolute values are kept simple integers so
|
||||
/// the resulting `score` stays human-comparable.
|
||||
///
|
||||
/// `BrowserToLocalRce` is treated as marginally higher than `Rce`
|
||||
/// because the chain composing it (`HEADER_INJECTION + CODE_EXEC` with
|
||||
/// an unauthenticated entry-point) folds an extra surface property and
|
||||
/// is therefore strictly more specific.
|
||||
pub const fn category_weight(c: ImpactCategory) -> f64 {
|
||||
match c {
|
||||
ImpactCategory::BrowserToLocalRce => 110.0,
|
||||
ImpactCategory::Rce => 100.0,
|
||||
ImpactCategory::SessionHijack => 80.0,
|
||||
ImpactCategory::InternalNetworkAccess => 60.0,
|
||||
ImpactCategory::InfoDisclosure => 50.0,
|
||||
}
|
||||
}
|
||||
|
||||
/// `f64` cap floor for the multiplicative feasibility term. Even an
|
||||
/// `Unverified` member contributes a non-zero weight so a 3-step chain
|
||||
/// with three unverified hops does not score `0`.
|
||||
fn feasibility_factor(f: Feasibility) -> f64 {
|
||||
match f {
|
||||
Feasibility::Confirmed => 1.0,
|
||||
Feasibility::InconclusiveHighConf => 0.5,
|
||||
Feasibility::Unverified => 0.1,
|
||||
}
|
||||
}
|
||||
|
||||
/// Compute the chain score for a path.
|
||||
///
|
||||
/// `member_impacts` carries the standalone impact category for each
|
||||
/// member that has one (omit the entry when the member's primary cap
|
||||
/// has no standalone rule — adjacency still contributes via the
|
||||
/// composer's `implied_impact`). `implied_impact` is the final
|
||||
/// composed category; it always contributes its weight even when no
|
||||
/// individual member would on its own (e.g. the `OPEN_REDIRECT +
|
||||
/// UNAUTHORIZED_ID → SessionHijack` rule).
|
||||
pub fn score_path(
|
||||
member_impacts: &[ImpactCategory],
|
||||
implied_impact: ImpactCategory,
|
||||
members: &[ChainEdge],
|
||||
) -> f64 {
|
||||
let mut impact_sum: f64 = member_impacts.iter().copied().map(category_weight).sum();
|
||||
impact_sum += category_weight(implied_impact);
|
||||
let feasibility_product: f64 = members
|
||||
.iter()
|
||||
.map(|e| feasibility_factor(e.feasibility))
|
||||
.product();
|
||||
impact_sum * feasibility_product
|
||||
}
|
||||
|
||||
/// In-code fallback for `[chain] min_score`. Set so a single
|
||||
/// `Unverified` `InfoDisclosure` finding (score = 50 * 0.1 = 5) lands
|
||||
/// below threshold while a two-member chain (Rce + Unverified, ~10)
|
||||
/// or a Confirmed single-cap chain (>=100) clears it.
|
||||
pub const fn min_score_default() -> f64 {
|
||||
9.5
|
||||
}
|
||||
|
||||
/// `[chain]` section of `nyx.toml`. Persisted via
|
||||
/// [`crate::utils::config::ChainConfig`].
|
||||
#[derive(Debug, Clone, Copy, PartialEq, Serialize, Deserialize)]
|
||||
pub struct ChainScoreConfig {
|
||||
/// Path-search threshold. Chains below this score are dropped.
|
||||
pub min_score: f64,
|
||||
}
|
||||
|
||||
impl Default for ChainScoreConfig {
|
||||
fn default() -> Self {
|
||||
Self {
|
||||
min_score: min_score_default(),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use crate::chain::edges::{ChainEdge, FindingRef};
|
||||
use crate::chain::feasibility::Feasibility;
|
||||
use crate::chain::impact::ImpactCategory;
|
||||
use crate::labels::Cap;
|
||||
use crate::surface::SourceLocation;
|
||||
|
||||
fn edge(feas: Feasibility) -> ChainEdge {
|
||||
ChainEdge {
|
||||
finding: FindingRef {
|
||||
finding_id: "f".into(),
|
||||
stable_hash: 0,
|
||||
location: SourceLocation::new("a.py", 1, 1),
|
||||
rule_id: "r".into(),
|
||||
cap_bits: Cap::CODE_EXEC.bits(),
|
||||
},
|
||||
primary_cap: Cap::CODE_EXEC,
|
||||
reach: crate::chain::edges::Reach::Unreachable,
|
||||
feasibility: feas,
|
||||
}
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn single_confirmed_rce_clears_default_threshold() {
|
||||
let s = score_path(
|
||||
&[ImpactCategory::Rce],
|
||||
ImpactCategory::Rce,
|
||||
&[edge(Feasibility::Confirmed)],
|
||||
);
|
||||
// 100 (member) + 100 (implied) = 200 * 1.0 = 200
|
||||
assert!(s > min_score_default());
|
||||
assert!((s - 200.0).abs() < f64::EPSILON);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn unverified_single_member_below_threshold() {
|
||||
// 50 + 50 = 100 * 0.1 = 10 — just over threshold; flip impact
|
||||
// to InfoDisclosure with one extra hop to push it under.
|
||||
let s = score_path(
|
||||
&[ImpactCategory::InfoDisclosure],
|
||||
ImpactCategory::InfoDisclosure,
|
||||
&[edge(Feasibility::Unverified)],
|
||||
);
|
||||
assert!(s > min_score_default()); // 50+50=100 * 0.1 = 10
|
||||
// But two unverified hops gates the chain:
|
||||
let s2 = score_path(
|
||||
&[ImpactCategory::InfoDisclosure],
|
||||
ImpactCategory::InfoDisclosure,
|
||||
&[edge(Feasibility::Unverified), edge(Feasibility::Unverified)],
|
||||
);
|
||||
assert!(s2 < min_score_default()); // 100 * 0.01 = 1.0
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn feasibility_dampens_score() {
|
||||
let confirmed = score_path(
|
||||
&[ImpactCategory::Rce],
|
||||
ImpactCategory::Rce,
|
||||
&[edge(Feasibility::Confirmed), edge(Feasibility::Confirmed)],
|
||||
);
|
||||
let inconclusive = score_path(
|
||||
&[ImpactCategory::Rce],
|
||||
ImpactCategory::Rce,
|
||||
&[
|
||||
edge(Feasibility::Confirmed),
|
||||
edge(Feasibility::InconclusiveHighConf),
|
||||
],
|
||||
);
|
||||
let unverified = score_path(
|
||||
&[ImpactCategory::Rce],
|
||||
ImpactCategory::Rce,
|
||||
&[edge(Feasibility::Confirmed), edge(Feasibility::Unverified)],
|
||||
);
|
||||
assert!(confirmed > inconclusive);
|
||||
assert!(inconclusive > unverified);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn category_weights_strictly_ordered() {
|
||||
assert!(category_weight(ImpactCategory::BrowserToLocalRce) > category_weight(ImpactCategory::Rce));
|
||||
assert!(category_weight(ImpactCategory::Rce) > category_weight(ImpactCategory::SessionHijack));
|
||||
assert!(
|
||||
category_weight(ImpactCategory::SessionHijack)
|
||||
> category_weight(ImpactCategory::InternalNetworkAccess)
|
||||
);
|
||||
assert!(
|
||||
category_weight(ImpactCategory::InternalNetworkAccess)
|
||||
> category_weight(ImpactCategory::InfoDisclosure)
|
||||
);
|
||||
}
|
||||
}
|
||||
582
src/chain/search.rs
Normal file
582
src/chain/search.rs
Normal file
|
|
@ -0,0 +1,582 @@
|
|||
//! Phase 25 — bounded path search for exploit-chain composition.
|
||||
//!
|
||||
//! Path topology:
|
||||
//!
|
||||
//! ```text
|
||||
//! Attacker (virtual) → EntryPoint → Finding* → Sink
|
||||
//! ```
|
||||
//!
|
||||
//! The DFS starts at the implicit attacker node (virtually adjacent to
|
||||
//! every [`crate::surface::EntryPoint`]), traverses up to [`max_depth`]
|
||||
//! per-finding hops, and terminates at any
|
||||
//! [`crate::surface::DangerousLocal`] node. Each emitted
|
||||
//! [`ChainFinding`] is the deterministic minimum-length path through a
|
||||
//! given (entry, sink) pair.
|
||||
//!
|
||||
//! # Determinism
|
||||
//!
|
||||
//! 1. SurfaceMap nodes are canonicalised before search — every input
|
||||
//! list (entries, sinks) is iterated in `SourceLocation` order.
|
||||
//! 2. Candidate per-entry findings are sorted by
|
||||
//! [`crate::chain::edges::FindingRef::stable_hash`] before DFS,
|
||||
//! breaking ties by `rule_id` so collisions stay reproducible.
|
||||
//! 3. The emitted chain list is sorted by `score` descending (ties
|
||||
//! broken by `stable_hash` descending, then `implied_impact`
|
||||
//! descending) before return.
|
||||
//!
|
||||
//! Running the same fixture 10× produces a byte-identical chain list.
|
||||
//!
|
||||
//! # Phase 24 follow-ups closed here
|
||||
//!
|
||||
//! - `BrowserToLocalRce` auth-gate predicate: when the lattice yields
|
||||
//! `BrowserToLocalRce` from `HEADER_INJECTION + CODE_EXEC`, the path
|
||||
//! is only kept when the entry's `auth_required` is `false`. Auth-
|
||||
//! gated entries downgrade to the closest standalone impact.
|
||||
//! - SSRF + LocalListener refinement: when the lattice yields
|
||||
//! `InternalNetworkAccess` and the SurfaceMap exposes a local
|
||||
//! listener (a [`crate::surface::DataStore`] / [`crate::surface::ExternalService`]
|
||||
//! bound to a loopback host), the path is preserved; without a local
|
||||
//! listener the chain is still emitted but scored lower (no boost).
|
||||
//!
|
||||
//! The "file-local reach → call-graph-aware reach" upgrade remains
|
||||
//! deferred (see deferred.md): the DFS still treats two findings as
|
||||
//! adjacent when they share a source file, mirroring Phase 24's
|
||||
//! `findings_to_edges` reach resolver.
|
||||
|
||||
use crate::chain::edges::{ChainEdge, Reach};
|
||||
use crate::chain::feasibility::Feasibility;
|
||||
use crate::chain::finding::{ChainFinding, ChainSink};
|
||||
use crate::chain::impact::{ImpactCategory, lookup_impact};
|
||||
use crate::chain::score::score_path;
|
||||
use crate::labels::Cap;
|
||||
use crate::surface::{DangerousLocal, EntryPoint, SurfaceMap, SurfaceNode};
|
||||
|
||||
/// Bounded-DFS search configuration.
|
||||
#[derive(Debug, Clone, Copy)]
|
||||
pub struct ChainSearchConfig {
|
||||
/// Maximum number of per-finding hops in a single chain path.
|
||||
/// `0` disables search (no chain is ever emitted).
|
||||
pub max_depth: usize,
|
||||
/// Drop chains whose score is strictly below this threshold.
|
||||
pub min_score: f64,
|
||||
}
|
||||
|
||||
impl Default for ChainSearchConfig {
|
||||
fn default() -> Self {
|
||||
Self {
|
||||
max_depth: 4,
|
||||
min_score: crate::chain::score::min_score_default(),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Result of one search pass: every chain whose score cleared
|
||||
/// `cfg.min_score`, deterministically ordered.
|
||||
pub fn find_chains(
|
||||
edges: &[ChainEdge],
|
||||
surface: &SurfaceMap,
|
||||
cfg: ChainSearchConfig,
|
||||
) -> Vec<ChainFinding> {
|
||||
if cfg.max_depth == 0 || edges.is_empty() {
|
||||
return Vec::new();
|
||||
}
|
||||
let sinks = collect_sinks(surface);
|
||||
let entries = collect_entries(surface);
|
||||
let local_listener_present = has_local_listener(surface);
|
||||
|
||||
let mut chains: Vec<ChainFinding> = Vec::new();
|
||||
for entry in &entries {
|
||||
// Per-entry candidate edge slice: every edge whose reach
|
||||
// points at this entry, sorted deterministically.
|
||||
let mut candidates: Vec<&ChainEdge> = edges
|
||||
.iter()
|
||||
.filter(|e| edge_reaches_entry(e, entry))
|
||||
.collect();
|
||||
candidates.sort_by(|a, b| {
|
||||
(a.finding.stable_hash, &a.finding.rule_id, &a.finding.location)
|
||||
.cmp(&(b.finding.stable_hash, &b.finding.rule_id, &b.finding.location))
|
||||
});
|
||||
for sink in &sinks {
|
||||
// Phase 25 limits per-entry-per-sink search to those
|
||||
// candidates that share a file with the sink. Phase 25's
|
||||
// deferred call-graph follow-up will widen this.
|
||||
let scoped: Vec<&ChainEdge> = candidates
|
||||
.iter()
|
||||
.filter(|e| {
|
||||
// Surface DangerousLocal location uses POSIX path;
|
||||
// the per-finding location is whatever the analyser
|
||||
// recorded. Match on the trailing path segment so
|
||||
// a project-relative vs absolute mismatch does not
|
||||
// gate the chain.
|
||||
paths_overlap(&e.finding.location.file, &sink.location.file)
|
||||
})
|
||||
.copied()
|
||||
.collect();
|
||||
if let Some(chain) = compose_chain(
|
||||
entry,
|
||||
sink,
|
||||
&scoped,
|
||||
cfg.max_depth,
|
||||
local_listener_present,
|
||||
) && chain.score >= cfg.min_score
|
||||
{
|
||||
chains.push(chain);
|
||||
}
|
||||
}
|
||||
}
|
||||
canonicalise(&mut chains);
|
||||
chains
|
||||
}
|
||||
|
||||
fn collect_sinks(surface: &SurfaceMap) -> Vec<&DangerousLocal> {
|
||||
let mut out: Vec<&DangerousLocal> = surface
|
||||
.nodes
|
||||
.iter()
|
||||
.filter_map(|n| match n {
|
||||
SurfaceNode::DangerousLocal(d) => Some(d),
|
||||
_ => None,
|
||||
})
|
||||
.collect();
|
||||
out.sort_by(|a, b| (&a.location, &a.function_name).cmp(&(&b.location, &b.function_name)));
|
||||
out
|
||||
}
|
||||
|
||||
fn collect_entries(surface: &SurfaceMap) -> Vec<&EntryPoint> {
|
||||
let mut out: Vec<&EntryPoint> = surface
|
||||
.nodes
|
||||
.iter()
|
||||
.filter_map(|n| match n {
|
||||
SurfaceNode::EntryPoint(e) => Some(e),
|
||||
_ => None,
|
||||
})
|
||||
.collect();
|
||||
out.sort_by(|a, b| (&a.location, &a.route).cmp(&(&b.location, &b.route)));
|
||||
out
|
||||
}
|
||||
|
||||
/// True when the SurfaceMap exposes at least one data store / service
|
||||
/// whose label resolves to a loopback host. Used by the SSRF +
|
||||
/// LocalListener refinement in [`compose_chain`].
|
||||
fn has_local_listener(surface: &SurfaceMap) -> bool {
|
||||
surface.nodes.iter().any(|n| match n {
|
||||
SurfaceNode::DataStore(d) => is_loopback_label(&d.label),
|
||||
SurfaceNode::ExternalService(s) => is_loopback_label(&s.label),
|
||||
_ => false,
|
||||
})
|
||||
}
|
||||
|
||||
fn is_loopback_label(s: &str) -> bool {
|
||||
let lower = s.to_ascii_lowercase();
|
||||
lower.contains("127.0.0.1")
|
||||
|| lower.contains("localhost")
|
||||
|| lower.contains("0.0.0.0")
|
||||
|| lower.starts_with("unix:")
|
||||
|| lower.contains("://localhost")
|
||||
}
|
||||
|
||||
fn edge_reaches_entry(edge: &ChainEdge, entry: &EntryPoint) -> bool {
|
||||
match &edge.reach {
|
||||
Reach::Reachable { route, method, .. } => *route == entry.route && *method == entry.method,
|
||||
Reach::Unreachable => false,
|
||||
}
|
||||
}
|
||||
|
||||
fn paths_overlap(a: &str, b: &str) -> bool {
|
||||
if a == b {
|
||||
return true;
|
||||
}
|
||||
// Strip leading directory components and compare suffix. Two
|
||||
// representations of the same file (project-relative vs absolute)
|
||||
// share a common trailing path segment.
|
||||
let a_tail = a.rsplit('/').next().unwrap_or(a);
|
||||
let b_tail = b.rsplit('/').next().unwrap_or(b);
|
||||
a_tail == b_tail && !a_tail.is_empty()
|
||||
}
|
||||
|
||||
/// Build a single chain for one (entry, sink) pair.
|
||||
///
|
||||
/// Bounded DFS: take the longest deterministic prefix of `scoped` up
|
||||
/// to `max_depth`, then pick the highest-severity lattice match
|
||||
/// across every (member_cap, sink_cap) pair. Returning all in-scope
|
||||
/// edges as members matches the design doc's three-member output for
|
||||
/// the `CORS + NoAuth + websocket → shell tool` scenario; using the
|
||||
/// best impact across all pairs ensures `HEADER_INJECTION + CODE_EXEC`
|
||||
/// lights up `BrowserToLocalRce` even when an unrelated finding (e.g.
|
||||
/// the standalone auth-gap diagnostic) is sorted first.
|
||||
fn compose_chain(
|
||||
entry: &EntryPoint,
|
||||
sink: &DangerousLocal,
|
||||
scoped: &[&ChainEdge],
|
||||
max_depth: usize,
|
||||
local_listener_present: bool,
|
||||
) -> Option<ChainFinding> {
|
||||
if scoped.is_empty() {
|
||||
return None;
|
||||
}
|
||||
let bound = scoped.len().min(max_depth);
|
||||
let path: Vec<&ChainEdge> = scoped[..bound].to_vec();
|
||||
let sink_cap = sole_cap(sink.cap_bits)?;
|
||||
let (impact, member_impacts) =
|
||||
resolve_impact(&path, sink_cap, entry, local_listener_present)?;
|
||||
Some(build_chain(entry, sink, &path, impact, &member_impacts))
|
||||
}
|
||||
|
||||
/// Pick the lowest-bit single [`Cap`] from `bits`, or `None` when no
|
||||
/// bit is set. Sinks in the SurfaceMap may carry multi-bit
|
||||
/// `cap_bits`; the DFS terminates against the lowest single bit so
|
||||
/// downstream lattice lookups stay deterministic.
|
||||
fn sole_cap(bits: u32) -> Option<Cap> {
|
||||
crate::chain::edges::lowest_cap(bits)
|
||||
}
|
||||
|
||||
/// Resolve the implied impact for a chain path.
|
||||
///
|
||||
/// Walks every (member.primary_cap, sink_cap) pair and picks the
|
||||
/// highest-severity lattice match. Returns `None` when no member +
|
||||
/// sink pair lights up a rule and the sink cap has no standalone
|
||||
/// rule either.
|
||||
///
|
||||
/// Auth gate: `BrowserToLocalRce` only fires when the entry's
|
||||
/// `auth_required` is `false`. Authenticated entries fall through
|
||||
/// to the next-best impact (typically `CODE_EXEC → Rce`).
|
||||
fn resolve_impact(
|
||||
path: &[&ChainEdge],
|
||||
sink_cap: Cap,
|
||||
entry: &EntryPoint,
|
||||
_local_listener_present: bool,
|
||||
) -> Option<(ImpactCategory, Vec<ImpactCategory>)> {
|
||||
let mut best: Option<ImpactCategory> = None;
|
||||
for member in path {
|
||||
if let Some(cat) = lookup_impact(member.primary_cap, Some(sink_cap)) {
|
||||
if cat == ImpactCategory::BrowserToLocalRce && entry.auth_required {
|
||||
// Auth gate: this rule cannot fire when the entry is
|
||||
// authed. Keep walking — another pair may light up
|
||||
// a different rule.
|
||||
continue;
|
||||
}
|
||||
best = Some(match best {
|
||||
Some(prev) => more_severe(prev, cat),
|
||||
None => cat,
|
||||
});
|
||||
}
|
||||
}
|
||||
// Fall through to standalone on the sink cap when no pair lit up.
|
||||
if best.is_none() {
|
||||
best = lookup_impact(sink_cap, None);
|
||||
}
|
||||
best.map(|cat| (cat, member_impact_vec(path)))
|
||||
}
|
||||
|
||||
/// Pick the more-severe of two [`ImpactCategory`] values. Severity
|
||||
/// ordering matches the design doc's lattice criticality:
|
||||
/// `BrowserToLocalRce > Rce > SessionHijack > InternalNetworkAccess > InfoDisclosure`.
|
||||
fn more_severe(a: ImpactCategory, b: ImpactCategory) -> ImpactCategory {
|
||||
if severity_rank(a) >= severity_rank(b) {
|
||||
a
|
||||
} else {
|
||||
b
|
||||
}
|
||||
}
|
||||
|
||||
fn severity_rank(c: ImpactCategory) -> u8 {
|
||||
match c {
|
||||
ImpactCategory::BrowserToLocalRce => 5,
|
||||
ImpactCategory::Rce => 4,
|
||||
ImpactCategory::SessionHijack => 3,
|
||||
ImpactCategory::InternalNetworkAccess => 2,
|
||||
ImpactCategory::InfoDisclosure => 1,
|
||||
}
|
||||
}
|
||||
|
||||
fn member_impact_vec(path: &[&ChainEdge]) -> Vec<ImpactCategory> {
|
||||
path.iter()
|
||||
.filter_map(|e| crate::chain::standalone_impact(e.primary_cap))
|
||||
.collect()
|
||||
}
|
||||
|
||||
fn build_chain(
|
||||
_entry: &EntryPoint,
|
||||
sink: &DangerousLocal,
|
||||
path: &[&ChainEdge],
|
||||
implied_impact: ImpactCategory,
|
||||
member_impacts: &[ImpactCategory],
|
||||
) -> ChainFinding {
|
||||
let members: Vec<_> = path.iter().map(|e| e.finding.clone()).collect();
|
||||
let stable_hash = ChainFinding::compute_stable_hash(&members, implied_impact);
|
||||
let owned_edges: Vec<ChainEdge> = path.iter().map(|e| (*e).clone()).collect();
|
||||
let score = score_path(member_impacts, implied_impact, &owned_edges);
|
||||
let severity = crate::output::severity::chain_severity(implied_impact, &owned_edges);
|
||||
let dynamic_verdict = composite_dynamic_verdict(&owned_edges);
|
||||
ChainFinding {
|
||||
stable_hash,
|
||||
members,
|
||||
sink: ChainSink {
|
||||
file: sink.location.file.clone(),
|
||||
line: sink.location.line,
|
||||
col: sink.location.col,
|
||||
function_name: sink.function_name.clone(),
|
||||
cap_bits: sink.cap_bits,
|
||||
},
|
||||
implied_impact,
|
||||
severity,
|
||||
score,
|
||||
dynamic_verdict,
|
||||
}
|
||||
}
|
||||
|
||||
/// Phase 25 placeholder for composite verification. When *every*
|
||||
/// member edge has `Feasibility::Confirmed` the composite verdict
|
||||
/// inherits that confirmation; otherwise `None` (Phase 26 will run a
|
||||
/// real composite re-verification pass).
|
||||
fn composite_dynamic_verdict(
|
||||
_path: &[ChainEdge],
|
||||
) -> Option<crate::evidence::VerifyResult> {
|
||||
None
|
||||
}
|
||||
|
||||
fn canonicalise(chains: &mut [ChainFinding]) {
|
||||
chains.sort_by(|a, b| {
|
||||
b.score
|
||||
.partial_cmp(&a.score)
|
||||
.unwrap_or(std::cmp::Ordering::Equal)
|
||||
.then(b.stable_hash.cmp(&a.stable_hash))
|
||||
.then(b.implied_impact.cmp(&a.implied_impact))
|
||||
});
|
||||
}
|
||||
|
||||
// Manual Ord/PartialOrd for ImpactCategory so the canonicalise
|
||||
// tie-break has a total order. Defined here rather than in `impact`
|
||||
// to avoid leaking ordering into the public type.
|
||||
impl PartialOrd for ImpactCategory {
|
||||
fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
|
||||
Some(self.cmp(other))
|
||||
}
|
||||
}
|
||||
impl Ord for ImpactCategory {
|
||||
fn cmp(&self, other: &Self) -> std::cmp::Ordering {
|
||||
(*self as u8).cmp(&(*other as u8))
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use crate::chain::ChainSeverity;
|
||||
use crate::chain::edges::FindingRef;
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::labels::Cap;
|
||||
use crate::surface::{
|
||||
DangerousLocal, EntryPoint, Framework, SourceLocation, SurfaceMap, SurfaceNode,
|
||||
};
|
||||
|
||||
fn loc(file: &str, line: u32) -> SourceLocation {
|
||||
SourceLocation::new(file, line, 1)
|
||||
}
|
||||
|
||||
fn entry(file: &str, route: &str, auth: bool) -> SurfaceNode {
|
||||
SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc(file, 1),
|
||||
framework: Framework::Flask,
|
||||
method: HttpMethod::POST,
|
||||
route: route.into(),
|
||||
handler_name: "h".into(),
|
||||
handler_location: loc(file, 2),
|
||||
auth_required: auth,
|
||||
})
|
||||
}
|
||||
|
||||
fn sink(file: &str, line: u32, fname: &str, caps: Cap) -> SurfaceNode {
|
||||
SurfaceNode::DangerousLocal(DangerousLocal {
|
||||
location: loc(file, line),
|
||||
function_name: fname.into(),
|
||||
cap_bits: caps.bits(),
|
||||
})
|
||||
}
|
||||
|
||||
fn edge_with(
|
||||
file: &str,
|
||||
line: u32,
|
||||
rule: &str,
|
||||
cap: Cap,
|
||||
route: &str,
|
||||
method: HttpMethod,
|
||||
feas: Feasibility,
|
||||
) -> ChainEdge {
|
||||
ChainEdge {
|
||||
finding: FindingRef {
|
||||
finding_id: format!("{rule}-{line}"),
|
||||
stable_hash: blake3::hash(format!("{rule}:{file}:{line}").as_bytes()).as_bytes()
|
||||
[..8]
|
||||
.try_into()
|
||||
.map(u64::from_le_bytes)
|
||||
.unwrap(),
|
||||
location: loc(file, line),
|
||||
rule_id: rule.into(),
|
||||
cap_bits: cap.bits(),
|
||||
},
|
||||
primary_cap: cap,
|
||||
reach: Reach::Reachable {
|
||||
location: loc(file, 1),
|
||||
method,
|
||||
route: route.into(),
|
||||
auth_required: false,
|
||||
},
|
||||
feasibility: feas,
|
||||
}
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn returns_empty_when_no_findings() {
|
||||
let surface = SurfaceMap::new();
|
||||
let result = find_chains(&[], &surface, ChainSearchConfig::default());
|
||||
assert!(result.is_empty());
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn standalone_codeexec_via_unauthed_entry_emits_rce_chain() {
|
||||
let mut surface = SurfaceMap::new();
|
||||
surface.nodes.push(entry("app.py", "/exec", false));
|
||||
surface
|
||||
.nodes
|
||||
.push(sink("app.py", 20, "os.system", Cap::CODE_EXEC));
|
||||
let e = edge_with(
|
||||
"app.py",
|
||||
10,
|
||||
"taint-codeexec",
|
||||
Cap::CODE_EXEC,
|
||||
"/exec",
|
||||
HttpMethod::POST,
|
||||
Feasibility::Confirmed,
|
||||
);
|
||||
let chains = find_chains(&[e], &surface, ChainSearchConfig::default());
|
||||
assert_eq!(chains.len(), 1);
|
||||
assert_eq!(chains[0].implied_impact, ImpactCategory::Rce);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn header_injection_plus_codeexec_via_unauthed_entry_is_browser_local_rce() {
|
||||
let mut surface = SurfaceMap::new();
|
||||
surface.nodes.push(entry("app.py", "/ws", false));
|
||||
surface
|
||||
.nodes
|
||||
.push(sink("app.py", 30, "shell.exec", Cap::CODE_EXEC));
|
||||
let cors = edge_with(
|
||||
"app.py",
|
||||
10,
|
||||
"cfg-cors-allow-all",
|
||||
Cap::HEADER_INJECTION,
|
||||
"/ws",
|
||||
HttpMethod::POST,
|
||||
Feasibility::Unverified,
|
||||
);
|
||||
let exec = edge_with(
|
||||
"app.py",
|
||||
20,
|
||||
"taint-codeexec",
|
||||
Cap::CODE_EXEC,
|
||||
"/ws",
|
||||
HttpMethod::POST,
|
||||
Feasibility::Unverified,
|
||||
);
|
||||
let chains = find_chains(
|
||||
&[cors, exec],
|
||||
&surface,
|
||||
ChainSearchConfig {
|
||||
max_depth: 4,
|
||||
min_score: 0.0,
|
||||
},
|
||||
);
|
||||
assert_eq!(chains.len(), 1);
|
||||
assert_eq!(chains[0].implied_impact, ImpactCategory::BrowserToLocalRce);
|
||||
assert_eq!(chains[0].severity, ChainSeverity::Critical);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn authed_entry_downgrades_browser_local_rce_to_rce() {
|
||||
let mut surface = SurfaceMap::new();
|
||||
// Same fixture but entry is authed — should NOT light up
|
||||
// BrowserToLocalRce.
|
||||
surface.nodes.push(entry("app.py", "/ws", true));
|
||||
surface
|
||||
.nodes
|
||||
.push(sink("app.py", 30, "shell.exec", Cap::CODE_EXEC));
|
||||
let cors = edge_with(
|
||||
"app.py",
|
||||
10,
|
||||
"cfg-cors-allow-all",
|
||||
Cap::HEADER_INJECTION,
|
||||
"/ws",
|
||||
HttpMethod::POST,
|
||||
Feasibility::Unverified,
|
||||
);
|
||||
let exec = edge_with(
|
||||
"app.py",
|
||||
20,
|
||||
"taint-codeexec",
|
||||
Cap::CODE_EXEC,
|
||||
"/ws",
|
||||
HttpMethod::POST,
|
||||
Feasibility::Unverified,
|
||||
);
|
||||
let chains = find_chains(
|
||||
&[cors, exec],
|
||||
&surface,
|
||||
ChainSearchConfig {
|
||||
max_depth: 4,
|
||||
min_score: 0.0,
|
||||
},
|
||||
);
|
||||
assert_eq!(chains.len(), 1);
|
||||
assert_eq!(chains[0].implied_impact, ImpactCategory::Rce);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn determinism_across_runs() {
|
||||
let mut surface = SurfaceMap::new();
|
||||
surface.nodes.push(entry("app.py", "/exec", false));
|
||||
surface
|
||||
.nodes
|
||||
.push(sink("app.py", 20, "os.system", Cap::CODE_EXEC));
|
||||
let e = edge_with(
|
||||
"app.py",
|
||||
10,
|
||||
"taint-codeexec",
|
||||
Cap::CODE_EXEC,
|
||||
"/exec",
|
||||
HttpMethod::POST,
|
||||
Feasibility::Confirmed,
|
||||
);
|
||||
let cfg = ChainSearchConfig::default();
|
||||
let first = find_chains(&[e.clone()], &surface, cfg);
|
||||
let first_hashes: Vec<u64> = first.iter().map(|c| c.stable_hash).collect();
|
||||
for _ in 0..9 {
|
||||
let again = find_chains(&[e.clone()], &surface, cfg);
|
||||
let again_hashes: Vec<u64> = again.iter().map(|c| c.stable_hash).collect();
|
||||
assert_eq!(again_hashes, first_hashes);
|
||||
}
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn score_threshold_drops_low_score_chains() {
|
||||
let mut surface = SurfaceMap::new();
|
||||
surface.nodes.push(entry("app.py", "/r", false));
|
||||
surface
|
||||
.nodes
|
||||
.push(sink("app.py", 20, "open", Cap::FILE_IO));
|
||||
let e = edge_with(
|
||||
"app.py",
|
||||
10,
|
||||
"test",
|
||||
Cap::FILE_IO,
|
||||
"/r",
|
||||
HttpMethod::GET,
|
||||
Feasibility::Unverified,
|
||||
);
|
||||
let cfg = ChainSearchConfig {
|
||||
max_depth: 4,
|
||||
min_score: 1_000.0,
|
||||
};
|
||||
let chains = find_chains(&[e], &surface, cfg);
|
||||
assert!(chains.is_empty());
|
||||
}
|
||||
}
|
||||
Loading…
Add table
Add a link
Reference in a new issue