[pitboss] phase 16: Track B — Rust + C + C++ harness emitter shapes

This commit is contained in:
pitboss 2026-05-15 08:35:40 -05:00
parent bf62ae6b9f
commit 76087f931a
31 changed files with 1969 additions and 100 deletions

View file

@ -808,6 +808,165 @@ fn compute_php_lockfile_hash(workdir: &Path) -> String {
format!("{:016x}", u64::from_le_bytes(out.as_bytes()[..8].try_into().unwrap()))
}
// ── C build sandbox ───────────────────────────────────────────────────────────
/// Prepare a compiled C binary for `spec`.
///
/// Checks a build cache keyed on `(main.c + entry.c hash, "c", toolchain_id)`.
/// On a cache hit returns immediately; otherwise runs
/// `cc -O0 -g -o nyx_harness main.c` in `workdir`.
///
/// Build isolation is NOT yet implemented (deferred). `cc` runs on the host.
pub fn prepare_c(spec: &HarnessSpec, workdir: &Path) -> Result<BuildResult, BuildError> {
let source_hash = compute_c_source_hash(workdir);
let cache_path = build_cache_path(&source_hash, "c", &spec.toolchain_id)?;
let binary = cache_path.join("nyx_harness");
if binary.exists() {
return Ok(BuildResult {
venv_path: cache_path,
cache_hit: true,
duration: std::time::Duration::ZERO,
});
}
let start = std::time::Instant::now();
const MAX_ATTEMPTS: u32 = 2;
const BACKOFF: [u64; 2] = [1, 4];
let mut last_err = String::new();
for attempt in 0..MAX_ATTEMPTS {
if attempt > 0 {
std::thread::sleep(std::time::Duration::from_secs(BACKOFF[attempt as usize - 1]));
}
let _ = std::fs::remove_dir_all(&cache_path);
std::fs::create_dir_all(&cache_path)?;
match try_build_c_binary(workdir, &binary) {
Ok(()) => {
return Ok(BuildResult {
venv_path: cache_path,
cache_hit: false,
duration: start.elapsed(),
});
}
Err(e) => {
last_err = e;
let _ = std::fs::remove_file(&binary);
}
}
}
Err(BuildError::BuildFailed { stderr: last_err, attempts: MAX_ATTEMPTS })
}
fn try_build_c_binary(workdir: &Path, binary_dest: &Path) -> Result<(), String> {
let cc_bin = std::env::var("NYX_CC_BIN").unwrap_or_else(|_| "cc".to_owned());
let output = Command::new(&cc_bin)
.args(["-O0", "-g", "-o", binary_dest.to_str().unwrap_or("nyx_harness"), "main.c"])
.current_dir(workdir)
.env_clear()
.env("PATH", std::env::var("PATH").unwrap_or_default())
.env("HOME", std::env::var("HOME").unwrap_or_default())
.output()
.map_err(|e| format!("cc: {e}"))?;
if !output.status.success() {
return Err(String::from_utf8_lossy(&output.stderr).into_owned());
}
Ok(())
}
fn compute_c_source_hash(workdir: &Path) -> String {
let mut h = Hasher::new();
for fname in &["main.c", "entry.c", "Makefile"] {
if let Ok(content) = std::fs::read(workdir.join(fname)) {
h.update(fname.as_bytes());
h.update(&content);
}
}
let out = h.finalize();
format!("{:016x}", u64::from_le_bytes(out.as_bytes()[..8].try_into().unwrap()))
}
// ── C++ build sandbox ─────────────────────────────────────────────────────────
/// Prepare a compiled C++ binary for `spec`.
pub fn prepare_cpp(spec: &HarnessSpec, workdir: &Path) -> Result<BuildResult, BuildError> {
let source_hash = compute_cpp_source_hash(workdir);
let cache_path = build_cache_path(&source_hash, "cpp", &spec.toolchain_id)?;
let binary = cache_path.join("nyx_harness");
if binary.exists() {
return Ok(BuildResult {
venv_path: cache_path,
cache_hit: true,
duration: std::time::Duration::ZERO,
});
}
let start = std::time::Instant::now();
const MAX_ATTEMPTS: u32 = 2;
const BACKOFF: [u64; 2] = [1, 4];
let mut last_err = String::new();
for attempt in 0..MAX_ATTEMPTS {
if attempt > 0 {
std::thread::sleep(std::time::Duration::from_secs(BACKOFF[attempt as usize - 1]));
}
let _ = std::fs::remove_dir_all(&cache_path);
std::fs::create_dir_all(&cache_path)?;
match try_build_cpp_binary(workdir, &binary) {
Ok(()) => {
return Ok(BuildResult {
venv_path: cache_path,
cache_hit: false,
duration: start.elapsed(),
});
}
Err(e) => {
last_err = e;
let _ = std::fs::remove_file(&binary);
}
}
}
Err(BuildError::BuildFailed { stderr: last_err, attempts: MAX_ATTEMPTS })
}
fn try_build_cpp_binary(workdir: &Path, binary_dest: &Path) -> Result<(), String> {
let cxx_bin = std::env::var("NYX_CXX_BIN").unwrap_or_else(|_| {
// Prefer c++ which resolves to the system default compiler driver.
"c++".to_owned()
});
let output = Command::new(&cxx_bin)
.args(["-O0", "-g", "-std=c++17", "-o", binary_dest.to_str().unwrap_or("nyx_harness"), "main.cpp"])
.current_dir(workdir)
.env_clear()
.env("PATH", std::env::var("PATH").unwrap_or_default())
.env("HOME", std::env::var("HOME").unwrap_or_default())
.output()
.map_err(|e| format!("c++: {e}"))?;
if !output.status.success() {
return Err(String::from_utf8_lossy(&output.stderr).into_owned());
}
Ok(())
}
fn compute_cpp_source_hash(workdir: &Path) -> String {
let mut h = Hasher::new();
for fname in &["main.cpp", "entry.cpp", "CMakeLists.txt"] {
if let Ok(content) = std::fs::read(workdir.join(fname)) {
h.update(fname.as_bytes());
h.update(&content);
}
}
let out = h.finalize();
format!("{:016x}", u64::from_le_bytes(out.as_bytes()[..8].try_into().unwrap()))
}
// ── Docker-isolated build step functions ─────────────────────────────────────
//
// Each function runs the language's build tool inside a Docker container with

View file

@ -180,19 +180,22 @@ mod tests {
use crate::symbol::Lang;
#[test]
fn build_unsupported_lang_returns_err() {
// C is not supported (no emitter exists for it).
fn build_unsupported_entry_kind_returns_err() {
// The Python emitter advertises a specific entry-kind set; an
// unsupported entry kind short-circuits with
// [`UnsupportedReason::EntryKindUnsupported`] before any harness
// source is generated.
let spec = HarnessSpec {
finding_id: "0000000000000001".into(),
entry_file: "main.c".into(),
entry_name: "handleRequest".into(),
entry_kind: EntryKind::Function,
lang: Lang::C,
toolchain_id: "c-stable".into(),
entry_file: "src/app.py".into(),
entry_name: "handler".into(),
entry_kind: EntryKind::LibraryApi,
lang: Lang::Python,
toolchain_id: "python-3".into(),
payload_slot: PayloadSlot::Param(0),
expected_cap: Cap::SQL_QUERY,
constraint_hints: vec![],
sink_file: "main.c".into(),
sink_file: "src/app.py".into(),
sink_line: 5,
spec_hash: "0000000000000000".into(),
derivation: crate::dynamic::spec::SpecDerivationStrategy::FromFlowSteps,

View file

@ -1,22 +1,108 @@
//! C harness emitter (stub).
//! C harness emitter.
//!
//! No harness source is generated yet — `emit` returns
//! [`UnsupportedReason::LangUnsupported`]. The module exists so that
//! [`crate::dynamic::lang::entry_kinds_supported`] can advertise the entry
//! kinds Track B will deliver (Phase 16: `main(argc, argv)`,
//! `LLVMFuzzerTestOneInput`, free functions with `(const char*, size_t)` or
//! `(int, char**)` shapes) and so the verifier can surface
//! `Inconclusive(EntryKindUnsupported { … })` instead of dropping C findings.
//! Phase 16 (Track B Rust + C/C++ vertical) replaces the stub body with
//! dispatch over [`CShape`] — the cross product of [`EntryKind`] and a
//! lightweight per-file shape detector that inspects the entry file for
//! `main(int argc, char *argv[])`, libFuzzer's `LLVMFuzzerTestOneInput`,
//! and free functions with `(const char*, size_t)` signatures.
//!
//! Each shape emits a single `main.c` that:
//! 1. Reads the payload from `NYX_PAYLOAD` / `NYX_PAYLOAD_B64` env vars.
//! 2. `#include`s `entry.c` (the user's vulnerable code) and dispatches
//! via the per-shape adapter.
//!
//! Build step: `prepare_c()` in `build_sandbox.rs` runs
//! `cc -O0 -o nyx_harness main.c` in the workdir.
//!
//! File layout in workdir:
//! ```text
//! main.c ← harness entry point (generated, includes entry.c)
//! entry.c ← user entry source (copied from project)
//! Makefile ← optional, generated for reference
//! ```
//!
//! Payload slot support:
//! - `PayloadSlot::Param(0)` — pass payload as the first parameter (string
//! or `(buf, len)` pair depending on shape).
//! - `PayloadSlot::EnvVar(name)` — set env var before invoking entry.
//! - `PayloadSlot::Argv(n)` — `main(argc, argv)` shape: appended to argv.
use crate::dynamic::lang::{HarnessSource, LangEmitter};
use crate::dynamic::spec::{EntryKind, HarnessSpec};
use crate::dynamic::spec::{EntryKind, HarnessSpec, PayloadSlot};
use crate::evidence::UnsupportedReason;
use std::path::PathBuf;
/// Zero-sized [`LangEmitter`] handle for C.
pub struct CEmitter;
/// Entry kinds the C emitter intends to support once Phase 16 lands.
const SUPPORTED: &[EntryKind] = &[EntryKind::Function];
/// Entry kinds the C emitter understands after Phase 16.
///
/// `Function` covers free functions (libfuzzer-style + plain (const
/// char*, size_t)). `CliSubcommand` covers `main(argc, argv)`.
/// `LibraryApi` covers libFuzzer `LLVMFuzzerTestOneInput`.
const SUPPORTED: &[EntryKind] = &[
EntryKind::Function,
EntryKind::CliSubcommand,
EntryKind::LibraryApi,
];
// ── Phase 16: shape detector ─────────────────────────────────────────────────
/// Concrete per-file shape resolved by reading the entry source.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum CShape {
/// `int main(int argc, char *argv[])`. Harness embeds payload into
/// argv and calls `main(argc, argv)` directly.
MainArgv,
/// libFuzzer-style: `int LLVMFuzzerTestOneInput(const uint8_t *data,
/// size_t size)`. Harness invokes with `payload` bytes + length.
LibfuzzerEntry,
/// Free function with `(const char *, size_t)` or `(const char *)`
/// signature. Harness invokes directly.
FreeFn,
}
impl CShape {
/// Detect the shape from `(spec, source)`.
pub fn detect(spec: &HarnessSpec, source: &str) -> Self {
let entry = spec.entry_name.as_str();
let kind = spec.entry_kind;
let has_main_argv = (source.contains("int main(") || source.contains("int main ("))
&& (source.contains("argc") || source.contains("char *argv")
|| source.contains("char* argv") || source.contains("char **argv"));
let has_libfuzzer = source.contains("LLVMFuzzerTestOneInput") || entry == "LLVMFuzzerTestOneInput";
if has_libfuzzer {
return Self::LibfuzzerEntry;
}
if entry == "main" || has_main_argv {
return Self::MainArgv;
}
match kind {
EntryKind::CliSubcommand => Self::MainArgv,
EntryKind::LibraryApi => Self::LibfuzzerEntry,
_ => Self::FreeFn,
}
}
}
/// Public wrapper: detect the shape for a finalised `HarnessSpec`, reading
/// the entry file from disk.
pub fn detect_shape(spec: &HarnessSpec) -> CShape {
let src = read_entry_source(&spec.entry_file);
CShape::detect(spec, &src)
}
fn read_entry_source(entry_file: &str) -> String {
let candidates = [PathBuf::from(entry_file), PathBuf::from(".").join(entry_file)];
for path in &candidates {
if let Ok(s) = std::fs::read_to_string(path) {
return s;
}
}
String::new()
}
/// Source of the `__nyx_probe` shim for the (future) C harness (Phase 06 —
/// Track C.1). Variadic over `const char *` args; hand-rolled JSON keeps
@ -208,8 +294,8 @@ static void __nyx_install_crash_guard(const char *sink_callee) {
}
impl LangEmitter for CEmitter {
fn emit(&self, _spec: &HarnessSpec) -> Result<HarnessSource, UnsupportedReason> {
Err(UnsupportedReason::LangUnsupported)
fn emit(&self, spec: &HarnessSpec) -> Result<HarnessSource, UnsupportedReason> {
emit(spec)
}
fn entry_kinds_supported(&self) -> &'static [EntryKind] {
@ -218,18 +304,198 @@ impl LangEmitter for CEmitter {
fn entry_kind_hint(&self, attempted: EntryKind) -> String {
format!(
"c emitter is a stub; once Phase 16 (Track B Rust + C/C++ vertical) lands it will support {SUPPORTED:?} plus libFuzzer + main(argc, argv) shapes — attempted `EntryKind::{attempted}`"
"c emitter supports {SUPPORTED:?}; this finding's enclosing context is `EntryKind::{attempted}` — see Phase 16 shape dispatch (main / libFuzzer / free function)"
)
}
}
/// Emit a C harness for `spec`.
pub fn emit(spec: &HarnessSpec) -> Result<HarnessSource, UnsupportedReason> {
let shape = detect_shape(spec);
match (&spec.payload_slot, shape) {
(PayloadSlot::Param(0) | PayloadSlot::EnvVar(_), _) => {}
(PayloadSlot::Argv(_), CShape::MainArgv) => {}
_ => return Err(UnsupportedReason::PayloadSlotUnsupported),
}
let main_c = generate_main_c(spec, shape);
let makefile = generate_makefile();
Ok(HarnessSource {
source: main_c,
filename: "main.c".into(),
command: vec!["./nyx_harness".into()],
extra_files: vec![("Makefile".into(), makefile)],
entry_subpath: Some("entry.c".into()),
})
}
/// Generate the harness `main.c` for the resolved shape.
fn generate_main_c(spec: &HarnessSpec, shape: CShape) -> String {
let invocation = invoke_for_shape(spec, shape);
format!(
r#"/* Nyx dynamic harness — auto-generated, do not edit (Phase 16 — CShape::{shape:?}). */
#include <stddef.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
/* Forward declarations: the entry file is appended below via `#include`
* so the harness can call user-defined functions without a separate
* compilation unit. */
static char *nyx_payload(void);
#include "entry.c"
int main(int argc, char *argv[]) {{
(void)argc; (void)argv;
char *payload = nyx_payload();
if (!payload) payload = (char*)"";
{invocation}
/* Intentionally no free(payload): payload is either a strdup/b64_decode
* heap pointer or a string literal substituted above when allocation
* failed. free() on the literal is UB; the process exits immediately
* so the kernel reclaims the heap copy. */
return 0;
}}
/* Minimal base64 decoder (no external deps). */
static int nyx_b64_value(unsigned char c) {{
if (c >= 'A' && c <= 'Z') return c - 'A';
if (c >= 'a' && c <= 'z') return c - 'a' + 26;
if (c >= '0' && c <= '9') return c - '0' + 52;
if (c == '+') return 62;
if (c == '/') return 63;
return -1;
}}
static char *nyx_b64_decode(const char *in) {{
size_t n = strlen(in);
char *out = (char *)malloc(n + 1);
if (!out) return NULL;
size_t outi = 0;
int buf = 0, bits = 0;
for (size_t i = 0; i < n; ++i) {{
if (in[i] == '\n' || in[i] == '\r' || in[i] == '=') continue;
int v = nyx_b64_value((unsigned char)in[i]);
if (v < 0) {{ free(out); return NULL; }}
buf = (buf << 6) | v;
bits += 6;
if (bits >= 8) {{
bits -= 8;
out[outi++] = (char)((buf >> bits) & 0xFF);
}}
}}
out[outi] = '\0';
return out;
}}
static char *nyx_payload(void) {{
const char *v = getenv("NYX_PAYLOAD");
if (v && *v) {{
return strdup(v);
}}
const char *b64 = getenv("NYX_PAYLOAD_B64");
if (b64 && *b64) {{
return nyx_b64_decode(b64);
}}
return strdup("");
}}
"#,
shape = shape,
invocation = invocation,
)
}
fn invoke_for_shape(spec: &HarnessSpec, shape: CShape) -> String {
let entry_fn = &spec.entry_name;
match shape {
CShape::FreeFn => match &spec.payload_slot {
PayloadSlot::EnvVar(name) => format!(
" setenv({name:?}, payload, 1);\n {entry_fn}(payload, strlen(payload));\n",
),
_ => format!(" {entry_fn}(payload, strlen(payload));\n"),
},
CShape::LibfuzzerEntry => {
// libFuzzer: `int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size)`.
format!(
" {entry_fn}((const uint8_t *)payload, strlen(payload));\n",
entry_fn = entry_fn,
)
}
CShape::MainArgv => {
// Rename the user-supplied entry to `nyx_entry_main` via macro so
// it does not collide with the harness `main` symbol when the
// entry source defines `int main(...)`. Fixture authors should
// expose the entry as a function named in `spec.entry_name`.
let pad = match &spec.payload_slot {
PayloadSlot::Argv(n) => *n,
_ => 0,
};
let mut buf = String::from(" char *new_argv[8];\n");
buf.push_str(" int new_argc = 0;\n");
buf.push_str(" new_argv[new_argc++] = (char*)\"nyx_harness\";\n");
for _ in 0..pad {
buf.push_str(" new_argv[new_argc++] = (char*)\"\";\n");
}
buf.push_str(" new_argv[new_argc++] = payload;\n");
buf.push_str(" new_argv[new_argc] = NULL;\n");
buf.push_str(&format!(" {entry_fn}(new_argc, new_argv);\n"));
buf
}
}
}
fn generate_makefile() -> String {
r#"# Phase 16 — reference Makefile, not used by the runner (the build sandbox
# calls cc directly). Kept so reproductions can re-build the harness by hand.
CC ?= cc
CFLAGS ?= -O0 -g
all: nyx_harness
nyx_harness: main.c entry.c
$(CC) $(CFLAGS) -o nyx_harness main.c
clean:
rm -f nyx_harness
"#
.to_owned()
}
#[cfg(test)]
mod tests {
use super::*;
use crate::dynamic::spec::{EntryKind, HarnessSpec, PayloadSlot};
use crate::labels::Cap;
use crate::symbol::Lang;
fn make_spec(payload_slot: PayloadSlot) -> HarnessSpec {
HarnessSpec {
finding_id: "c00000000000001".into(),
entry_file: "entry.c".into(),
entry_name: "run".into(),
entry_kind: EntryKind::Function,
lang: Lang::C,
toolchain_id: "gcc-stable".into(),
payload_slot,
expected_cap: Cap::CODE_EXEC,
constraint_hints: vec![],
sink_file: "entry.c".into(),
sink_line: 10,
spec_hash: "ctest0000000001".into(),
derivation: crate::dynamic::spec::SpecDerivationStrategy::FromFlowSteps,
stubs_required: vec![],
}
}
#[test]
fn entry_kinds_supported_is_non_empty() {
assert!(!CEmitter.entry_kinds_supported().is_empty());
assert!(CEmitter.entry_kinds_supported().contains(&EntryKind::Function));
assert!(CEmitter.entry_kinds_supported().contains(&EntryKind::CliSubcommand));
assert!(CEmitter.entry_kinds_supported().contains(&EntryKind::LibraryApi));
}
#[test]
@ -238,4 +504,67 @@ mod tests {
assert!(hint.contains("LibraryApi"));
assert!(hint.contains("Phase 16"));
}
#[test]
fn shape_detect_main_argv() {
let src = "int main(int argc, char *argv[]) { return 0; }";
let mut spec = make_spec(PayloadSlot::Argv(0));
spec.entry_kind = EntryKind::CliSubcommand;
spec.entry_name = "main".into();
assert_eq!(CShape::detect(&spec, src), CShape::MainArgv);
}
#[test]
fn shape_detect_libfuzzer_entry() {
let src = "int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) { return 0; }";
let mut spec = make_spec(PayloadSlot::Param(0));
spec.entry_kind = EntryKind::LibraryApi;
spec.entry_name = "LLVMFuzzerTestOneInput".into();
assert_eq!(CShape::detect(&spec, src), CShape::LibfuzzerEntry);
}
#[test]
fn shape_detect_free_fn() {
let src = "void run(const char *s, size_t n) { (void)s; (void)n; }";
let spec = make_spec(PayloadSlot::Param(0));
assert_eq!(CShape::detect(&spec, src), CShape::FreeFn);
}
#[test]
fn emit_produces_source() {
let spec = make_spec(PayloadSlot::Param(0));
let h = emit(&spec).unwrap();
assert_eq!(h.filename, "main.c");
assert!(h.source.contains("#include \"entry.c\""));
assert!(h.source.contains("run(payload, strlen(payload))"));
assert_eq!(h.command, vec!["./nyx_harness"]);
assert_eq!(h.entry_subpath, Some("entry.c".to_string()));
}
#[test]
fn emit_main_argv_shape_routes_through_new_argv() {
let mut spec = make_spec(PayloadSlot::Argv(0));
spec.entry_kind = EntryKind::CliSubcommand;
spec.entry_name = "nyx_entry_main".into();
let h = emit(&spec).unwrap();
assert!(h.source.contains("new_argv[new_argc++] = payload"));
assert!(h.source.contains("nyx_entry_main(new_argc, new_argv)"));
}
#[test]
fn emit_libfuzzer_shape_passes_bytes() {
let mut spec = make_spec(PayloadSlot::Param(0));
spec.entry_kind = EntryKind::LibraryApi;
spec.entry_name = "LLVMFuzzerTestOneInput".into();
let h = emit(&spec).unwrap();
assert!(h.source.contains("LLVMFuzzerTestOneInput((const uint8_t *)payload, strlen(payload))"));
}
#[test]
fn emit_makefile_in_extra_files() {
let spec = make_spec(PayloadSlot::Param(0));
let h = emit(&spec).unwrap();
let mk = h.extra_files.iter().find(|(n, _)| n == "Makefile").expect("Makefile must be staged");
assert!(mk.1.contains("nyx_harness: main.c entry.c"));
}
}

View file

@ -1,22 +1,88 @@
//! C++ harness emitter (stub).
//! C++ harness emitter.
//!
//! No harness source is generated yet — `emit` returns
//! [`UnsupportedReason::LangUnsupported`]. The module exists so that
//! [`crate::dynamic::lang::entry_kinds_supported`] can advertise the entry
//! kinds Track B will deliver (Phase 16: `main(argc, argv)`,
//! `LLVMFuzzerTestOneInput`, free functions with `(const char*, size_t)`)
//! and so the verifier can surface `Inconclusive(EntryKindUnsupported { … })`
//! instead of dropping C++ findings.
//! Phase 16 (Track B Rust + C/C++ vertical) replaces the stub body with
//! dispatch over [`CppShape`] — `main(int argc, char *argv[])`, libFuzzer
//! `LLVMFuzzerTestOneInput`, and free functions with `(const char*,
//! size_t)` or `(const std::string&)` signatures.
//!
//! File layout in workdir:
//! ```text
//! main.cpp ← harness entry point (generated, includes entry.cpp)
//! entry.cpp ← user entry source (copied from project)
//! CMakeLists.txt ← optional, generated for reference
//! ```
//!
//! Build step: `prepare_cpp()` in `build_sandbox.rs` runs
//! `g++ -O0 -std=c++17 -o nyx_harness main.cpp` in the workdir.
use crate::dynamic::lang::{HarnessSource, LangEmitter};
use crate::dynamic::spec::{EntryKind, HarnessSpec};
use crate::dynamic::spec::{EntryKind, HarnessSpec, PayloadSlot};
use crate::evidence::UnsupportedReason;
use std::path::PathBuf;
/// Zero-sized [`LangEmitter`] handle for C++.
pub struct CppEmitter;
/// Entry kinds the C++ emitter intends to support once Phase 16 lands.
const SUPPORTED: &[EntryKind] = &[EntryKind::Function];
/// Entry kinds the C++ emitter understands after Phase 16.
const SUPPORTED: &[EntryKind] = &[
EntryKind::Function,
EntryKind::CliSubcommand,
EntryKind::LibraryApi,
];
// ── Phase 16: shape detector ─────────────────────────────────────────────────
/// Concrete per-file shape resolved by reading the entry source.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum CppShape {
/// `int main(int argc, char *argv[])`.
MainArgv,
/// libFuzzer-style: `int LLVMFuzzerTestOneInput(const uint8_t *, size_t)`.
LibfuzzerEntry,
/// Free function with `(const char *, size_t)` or `(const std::string&)`
/// signature.
FreeFn,
}
impl CppShape {
pub fn detect(spec: &HarnessSpec, source: &str) -> Self {
let entry = spec.entry_name.as_str();
let kind = spec.entry_kind;
let has_main_argv = (source.contains("int main(") || source.contains("int main ("))
&& (source.contains("argc") || source.contains("char *argv")
|| source.contains("char* argv") || source.contains("char **argv"));
let has_libfuzzer = source.contains("LLVMFuzzerTestOneInput")
|| entry == "LLVMFuzzerTestOneInput";
if has_libfuzzer {
return Self::LibfuzzerEntry;
}
if entry == "main" || has_main_argv {
return Self::MainArgv;
}
match kind {
EntryKind::CliSubcommand => Self::MainArgv,
EntryKind::LibraryApi => Self::LibfuzzerEntry,
_ => Self::FreeFn,
}
}
}
pub fn detect_shape(spec: &HarnessSpec) -> CppShape {
let src = read_entry_source(&spec.entry_file);
CppShape::detect(spec, &src)
}
fn read_entry_source(entry_file: &str) -> String {
let candidates = [PathBuf::from(entry_file), PathBuf::from(".").join(entry_file)];
for path in &candidates {
if let Ok(s) = std::fs::read_to_string(path) {
return s;
}
}
String::new()
}
/// Source of the `__nyx_probe` shim for the (future) C++ harness
/// (Phase 06 — Track C.1). Uses `<fstream>` + variadic templates; the
@ -201,8 +267,8 @@ inline void __nyx_install_crash_guard(const char *sink_callee) {
}
impl LangEmitter for CppEmitter {
fn emit(&self, _spec: &HarnessSpec) -> Result<HarnessSource, UnsupportedReason> {
Err(UnsupportedReason::LangUnsupported)
fn emit(&self, spec: &HarnessSpec) -> Result<HarnessSource, UnsupportedReason> {
emit(spec)
}
fn entry_kinds_supported(&self) -> &'static [EntryKind] {
@ -211,18 +277,182 @@ impl LangEmitter for CppEmitter {
fn entry_kind_hint(&self, attempted: EntryKind) -> String {
format!(
"cpp emitter is a stub; once Phase 16 (Track B Rust + C/C++ vertical) lands it will support {SUPPORTED:?} plus libFuzzer + main(argc, argv) shapes — attempted `EntryKind::{attempted}`"
"cpp emitter supports {SUPPORTED:?}; this finding's enclosing context is `EntryKind::{attempted}` — see Phase 16 shape dispatch (main / libFuzzer / free function)"
)
}
}
/// Emit a C++ harness for `spec`.
pub fn emit(spec: &HarnessSpec) -> Result<HarnessSource, UnsupportedReason> {
let shape = detect_shape(spec);
match (&spec.payload_slot, shape) {
(PayloadSlot::Param(0) | PayloadSlot::EnvVar(_), _) => {}
(PayloadSlot::Argv(_), CppShape::MainArgv) => {}
_ => return Err(UnsupportedReason::PayloadSlotUnsupported),
}
let main_cpp = generate_main_cpp(spec, shape);
let cmake = generate_cmake();
Ok(HarnessSource {
source: main_cpp,
filename: "main.cpp".into(),
command: vec!["./nyx_harness".into()],
extra_files: vec![("CMakeLists.txt".into(), cmake)],
entry_subpath: Some("entry.cpp".into()),
})
}
fn generate_main_cpp(spec: &HarnessSpec, shape: CppShape) -> String {
let invocation = invoke_for_shape(spec, shape);
format!(
r#"// Nyx dynamic harness — auto-generated, do not edit (Phase 16 — CppShape::{shape:?}).
#include <cstddef>
#include <cstdint>
#include <cstdlib>
#include <cstring>
#include <string>
#include <vector>
#include <iostream>
static std::string nyx_payload();
#include "entry.cpp"
int main(int argc, char *argv[]) {{
(void)argc; (void)argv;
std::string payload = nyx_payload();
{invocation}
return 0;
}}
// Minimal base64 decoder (no external deps).
static int nyx_b64_value(unsigned char c) {{
if (c >= 'A' && c <= 'Z') return c - 'A';
if (c >= 'a' && c <= 'z') return c - 'a' + 26;
if (c >= '0' && c <= '9') return c - '0' + 52;
if (c == '+') return 62;
if (c == '/') return 63;
return -1;
}}
static std::string nyx_b64_decode(const std::string &in) {{
std::string out;
int buf = 0, bits = 0;
for (char c : in) {{
if (c == '\n' || c == '\r' || c == '=') continue;
int v = nyx_b64_value(static_cast<unsigned char>(c));
if (v < 0) return std::string();
buf = (buf << 6) | v;
bits += 6;
if (bits >= 8) {{
bits -= 8;
out.push_back(static_cast<char>((buf >> bits) & 0xFF));
}}
}}
return out;
}}
static std::string nyx_payload() {{
if (const char *v = std::getenv("NYX_PAYLOAD")) {{
if (*v) return std::string(v);
}}
if (const char *b64 = std::getenv("NYX_PAYLOAD_B64")) {{
if (*b64) return nyx_b64_decode(std::string(b64));
}}
return std::string();
}}
"#,
shape = shape,
invocation = invocation,
)
}
fn invoke_for_shape(spec: &HarnessSpec, shape: CppShape) -> String {
let entry_fn = &spec.entry_name;
match shape {
CppShape::FreeFn => match &spec.payload_slot {
PayloadSlot::EnvVar(name) => format!(
" setenv({name:?}, payload.c_str(), 1);\n {entry_fn}(payload.c_str(), payload.size());\n",
),
_ => format!(" {entry_fn}(payload.c_str(), payload.size());\n"),
},
CppShape::LibfuzzerEntry => {
format!(
" {entry_fn}(reinterpret_cast<const uint8_t*>(payload.data()), payload.size());\n",
entry_fn = entry_fn,
)
}
CppShape::MainArgv => {
let pad = match &spec.payload_slot {
PayloadSlot::Argv(n) => *n,
_ => 0,
};
let mut buf = String::from(" std::vector<char*> new_argv;\n");
buf.push_str(" std::vector<std::string> argv_storage;\n");
buf.push_str(" argv_storage.emplace_back(\"nyx_harness\");\n");
for _ in 0..pad {
buf.push_str(" argv_storage.emplace_back(\"\");\n");
}
buf.push_str(" argv_storage.push_back(payload);\n");
buf.push_str(" for (auto &s : argv_storage) new_argv.push_back(s.data());\n");
buf.push_str(" new_argv.push_back(nullptr);\n");
buf.push_str(&format!(
" {entry_fn}(static_cast<int>(argv_storage.size()), new_argv.data());\n",
));
buf
}
}
}
fn generate_cmake() -> String {
r#"# Phase 16 — reference CMakeLists.txt, not used by the runner (the build
# sandbox calls g++ / clang++ directly). Kept so reproductions can re-build
# the harness by hand via `cmake -B build && cmake --build build`.
cmake_minimum_required(VERSION 3.10)
project(nyx_harness CXX)
set(CMAKE_CXX_STANDARD 17)
set(CMAKE_CXX_STANDARD_REQUIRED ON)
add_executable(nyx_harness main.cpp)
"#
.to_owned()
}
#[cfg(test)]
mod tests {
use super::*;
use crate::dynamic::spec::{EntryKind, HarnessSpec, PayloadSlot};
use crate::labels::Cap;
use crate::symbol::Lang;
fn make_spec(payload_slot: PayloadSlot) -> HarnessSpec {
HarnessSpec {
finding_id: "cpp0000000000001".into(),
entry_file: "entry.cpp".into(),
entry_name: "run".into(),
entry_kind: EntryKind::Function,
lang: Lang::Cpp,
toolchain_id: "g++-stable".into(),
payload_slot,
expected_cap: Cap::CODE_EXEC,
constraint_hints: vec![],
sink_file: "entry.cpp".into(),
sink_line: 10,
spec_hash: "cpptest00000001".into(),
derivation: crate::dynamic::spec::SpecDerivationStrategy::FromFlowSteps,
stubs_required: vec![],
}
}
#[test]
fn entry_kinds_supported_is_non_empty() {
assert!(!CppEmitter.entry_kinds_supported().is_empty());
assert!(CppEmitter.entry_kinds_supported().contains(&EntryKind::Function));
assert!(CppEmitter.entry_kinds_supported().contains(&EntryKind::CliSubcommand));
assert!(CppEmitter.entry_kinds_supported().contains(&EntryKind::LibraryApi));
}
#[test]
@ -231,4 +461,67 @@ mod tests {
assert!(hint.contains("CliSubcommand"));
assert!(hint.contains("Phase 16"));
}
#[test]
fn shape_detect_main_argv() {
let src = "int main(int argc, char *argv[]) { return 0; }";
let mut spec = make_spec(PayloadSlot::Argv(0));
spec.entry_kind = EntryKind::CliSubcommand;
spec.entry_name = "main".into();
assert_eq!(CppShape::detect(&spec, src), CppShape::MainArgv);
}
#[test]
fn shape_detect_libfuzzer() {
let src = "extern \"C\" int LLVMFuzzerTestOneInput(const uint8_t* d, size_t n) { return 0; }";
let mut spec = make_spec(PayloadSlot::Param(0));
spec.entry_kind = EntryKind::LibraryApi;
spec.entry_name = "LLVMFuzzerTestOneInput".into();
assert_eq!(CppShape::detect(&spec, src), CppShape::LibfuzzerEntry);
}
#[test]
fn shape_detect_free_fn() {
let src = "void run(const char *s, size_t n) { (void)s; (void)n; }";
let spec = make_spec(PayloadSlot::Param(0));
assert_eq!(CppShape::detect(&spec, src), CppShape::FreeFn);
}
#[test]
fn emit_produces_source() {
let spec = make_spec(PayloadSlot::Param(0));
let h = emit(&spec).unwrap();
assert_eq!(h.filename, "main.cpp");
assert!(h.source.contains("#include \"entry.cpp\""));
assert!(h.source.contains("run(payload.c_str(), payload.size())"));
assert_eq!(h.command, vec!["./nyx_harness"]);
assert_eq!(h.entry_subpath, Some("entry.cpp".to_string()));
}
#[test]
fn emit_libfuzzer_shape_passes_bytes() {
let mut spec = make_spec(PayloadSlot::Param(0));
spec.entry_kind = EntryKind::LibraryApi;
spec.entry_name = "LLVMFuzzerTestOneInput".into();
let h = emit(&spec).unwrap();
assert!(h.source.contains("LLVMFuzzerTestOneInput(reinterpret_cast<const uint8_t*>(payload.data()), payload.size())"));
}
#[test]
fn emit_main_argv_shape_builds_argv() {
let mut spec = make_spec(PayloadSlot::Argv(0));
spec.entry_kind = EntryKind::CliSubcommand;
spec.entry_name = "nyx_entry_main".into();
let h = emit(&spec).unwrap();
assert!(h.source.contains("argv_storage.push_back(payload)"));
assert!(h.source.contains("nyx_entry_main(static_cast<int>(argv_storage.size()), new_argv.data())"));
}
#[test]
fn emit_cmake_in_extra_files() {
let spec = make_spec(PayloadSlot::Param(0));
let h = emit(&spec).unwrap();
let mk = h.extra_files.iter().find(|(n, _)| n == "CMakeLists.txt").expect("CMakeLists.txt must be staged");
assert!(mk.1.contains("add_executable(nyx_harness main.cpp)"));
}
}

View file

@ -26,15 +26,24 @@ use crate::dynamic::lang::{HarnessSource, LangEmitter};
use crate::dynamic::spec::{EntryKind, HarnessSpec, PayloadSlot};
use crate::evidence::UnsupportedReason;
use crate::labels::Cap;
use std::path::PathBuf;
/// Zero-sized [`LangEmitter`] handle for Rust. Method bodies delegate to the
/// existing free functions in this module.
pub struct RustEmitter;
/// Entry kinds the Rust emitter currently understands. Extended in Phase 16
/// (Track B Rust + C/C++ vertical) to include `HttpRoute` (`actix_web`,
/// `axum`), `CliSubcommand` (clap), and `LibraryApi` (libfuzzer).
const SUPPORTED: &[EntryKind] = &[EntryKind::Function];
/// Entry kinds the Rust emitter understands after Phase 16.
///
/// `HttpRoute` covers `actix_web` and `axum` handlers. `CliSubcommand`
/// covers clap-driven CLIs. `LibraryApi` covers libfuzzer
/// `fuzz_target!` entry points. `Function` covers plain free functions
/// and is the fallback when shape detection is inconclusive.
const SUPPORTED: &[EntryKind] = &[
EntryKind::Function,
EntryKind::HttpRoute,
EntryKind::CliSubcommand,
EntryKind::LibraryApi,
];
impl LangEmitter for RustEmitter {
fn emit(&self, spec: &HarnessSpec) -> Result<HarnessSource, UnsupportedReason> {
@ -47,7 +56,7 @@ impl LangEmitter for RustEmitter {
fn entry_kind_hint(&self, attempted: EntryKind) -> String {
format!(
"rust emitter supports {SUPPORTED:?}; this finding's enclosing context is `EntryKind::{attempted}` — Track B will add actix / axum / clap / libfuzzer shapes in phase 16"
"rust emitter supports {SUPPORTED:?}; this finding's enclosing context is `EntryKind::{attempted}` — see Phase 16 shape dispatch (actix / axum / clap / libfuzzer)"
)
}
@ -303,15 +312,117 @@ fn __nyx_install_crash_guard(_sink_callee: &'static str) {}
"#
}
// ── Phase 16: shape detector ─────────────────────────────────────────────────
/// Concrete per-file shape resolved by reading the entry source.
///
/// One harness template per variant. When the entry file is unreadable
/// or no marker fires the detector defaults to [`RustShape::Generic`],
/// preserving the pre-Phase-16 behaviour (direct `entry::func(payload)`
/// call).
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum RustShape {
/// `actix_web` handler — `async fn handler(req: HttpRequest) -> HttpResponse`
/// or similar. Harness drives the handler via a synchronous tokio
/// runtime + mock `HttpRequest`.
ActixWebRoute,
/// `axum` handler — `async fn handler(...) -> impl IntoResponse`.
/// Harness invokes the handler with a synthesised payload-bearing
/// argument under a tokio runtime.
AxumHandler,
/// clap-driven CLI: `entry` parses `std::env::args` via `clap`.
/// Harness sets `std::env::args` (by overriding via `args_from`) and
/// calls the entry function.
ClapCli,
/// libfuzzer target — `fuzz_target!(|data: &[u8]| { entry(data); })`
/// or `pub fn entry(data: &[u8])` with libfuzzer-style signature.
/// Harness invokes with `payload.as_bytes()`.
LibfuzzerTarget,
/// Plain free function — `fn entry(payload: &str)`. Pre-Phase-16 default.
Generic,
}
impl RustShape {
/// Detect the shape from `(spec, source)`. `source` is the literal
/// bytes of the entry file (best-effort — empty string falls back
/// to [`Self::Generic`]).
pub fn detect(spec: &HarnessSpec, source: &str) -> Self {
let kind = spec.entry_kind;
let entry = spec.entry_name.as_str();
let has_actix = source.contains("actix_web::")
|| source.contains("HttpRequest")
|| source.contains("HttpResponse")
|| source.contains("#[get(")
|| source.contains("#[post(");
let has_axum = source.contains("axum::")
|| source.contains("IntoResponse")
|| source.contains("Json(")
|| source.contains("Query(")
|| source.contains("axum::extract");
let has_clap = source.contains("clap::")
|| source.contains("#[derive(Parser)")
|| source.contains("Parser::parse");
let has_libfuzzer = source.contains("libfuzzer_sys::fuzz_target")
|| source.contains("fuzz_target!")
|| (source.contains("pub fn ") && source.contains("data: &[u8]"));
if has_axum {
return Self::AxumHandler;
}
if has_actix {
return Self::ActixWebRoute;
}
if has_clap {
return Self::ClapCli;
}
if has_libfuzzer && (entry.starts_with("fuzz") || entry == "fuzz_target") {
return Self::LibfuzzerTarget;
}
match kind {
EntryKind::HttpRoute => Self::ActixWebRoute,
EntryKind::CliSubcommand => Self::ClapCli,
EntryKind::LibraryApi => Self::LibfuzzerTarget,
_ => Self::Generic,
}
}
}
/// Public wrapper to detect the shape for a finalised `HarnessSpec`,
/// reading the entry file from disk.
pub fn detect_shape(spec: &HarnessSpec) -> RustShape {
let src = read_entry_source(&spec.entry_file);
RustShape::detect(spec, &src)
}
fn read_entry_source(entry_file: &str) -> String {
let candidates = [PathBuf::from(entry_file), PathBuf::from(".").join(entry_file)];
for path in &candidates {
if let Ok(s) = std::fs::read_to_string(path) {
return s;
}
}
String::new()
}
/// Emit a Rust harness for `spec`.
pub fn emit(spec: &HarnessSpec) -> Result<HarnessSource, UnsupportedReason> {
match &spec.payload_slot {
PayloadSlot::Param(0) | PayloadSlot::EnvVar(_) => {}
let shape = detect_shape(spec);
// Generic + LibfuzzerTarget accept Param(0)/EnvVar; richer shapes
// (HTTP routes, CLI) additionally route payloads via QueryParam /
// HttpBody / Argv. Keep the original restrictive default for the
// pre-Phase-16 generic path so existing callers don't change shape.
match (&spec.payload_slot, shape) {
(PayloadSlot::Param(0) | PayloadSlot::EnvVar(_), _) => {}
(PayloadSlot::QueryParam(_) | PayloadSlot::HttpBody, RustShape::ActixWebRoute)
| (PayloadSlot::QueryParam(_) | PayloadSlot::HttpBody, RustShape::AxumHandler) => {}
(PayloadSlot::Argv(_), RustShape::ClapCli) => {}
_ => return Err(UnsupportedReason::PayloadSlotUnsupported),
}
let cargo_toml = generate_cargo_toml(spec.expected_cap);
let main_rs = generate_main_rs(spec);
let main_rs = generate_main_rs(spec, shape);
Ok(HarnessSource {
source: main_rs,
@ -350,17 +461,18 @@ pub fn generate_cargo_toml(cap: Cap) -> String {
/// Generate `src/main.rs` — the harness entry point.
///
/// Reads the payload from env, calls `entry::{entry_name}` with the payload
/// routed according to `spec.payload_slot`.
fn generate_main_rs(spec: &HarnessSpec) -> String {
/// routed according to `spec.payload_slot` and `shape`.
fn generate_main_rs(spec: &HarnessSpec, shape: RustShape) -> String {
let entry_fn = &spec.entry_name;
let (pre_call, call_expr) = build_call(spec, entry_fn);
let (pre_call, call_expr) = build_call(spec, entry_fn, shape);
format!(
r#"//! Nyx dynamic harness — auto-generated, do not edit.
r#"//! Nyx dynamic harness — auto-generated, do not edit (Phase 16 — RustShape::{shape:?}).
mod entry;
fn main() {{
let payload = nyx_payload();
let _ = &payload;
{pre_call} {call_expr}
}}
@ -412,33 +524,78 @@ fn b64_decode(input: &[u8]) -> Option<Vec<u8>> {{
Some(out)
}}
"#,
shape = shape,
pre_call = pre_call,
call_expr = call_expr,
)
}
/// Build `(pre_call_setup, call_expression)` strings for the chosen payload slot.
fn build_call(spec: &HarnessSpec, func: &str) -> (String, String) {
match &spec.payload_slot {
PayloadSlot::Param(0) => {
let pre = String::new();
let call = format!("entry::{func}(&payload);");
(pre, call)
}
PayloadSlot::EnvVar(name) => {
let pre = format!(" std::env::set_var({name:?}, &payload);\n");
let call = format!("entry::{func}();");
(pre, call)
}
_ => {
// Unreachable: `emit()` rejects all other slots up front.
let pre = String::new();
let call = format!("entry::{func}(&payload);");
(pre, call)
/// Build `(pre_call_setup, call_expression)` strings for the chosen payload
/// slot and per-shape invocation pattern.
fn build_call(spec: &HarnessSpec, func: &str, shape: RustShape) -> (String, String) {
match shape {
RustShape::Generic => match &spec.payload_slot {
PayloadSlot::Param(0) => (String::new(), format!("entry::{func}(&payload);")),
PayloadSlot::EnvVar(name) => (
format!(" std::env::set_var({name:?}, &payload);\n"),
format!("entry::{func}();"),
),
_ => (String::new(), format!("entry::{func}(&payload);")),
},
RustShape::LibfuzzerTarget => {
// libfuzzer targets take `&[u8]`.
(String::new(), format!("entry::{func}(payload.as_bytes());"))
}
RustShape::ActixWebRoute => actix_invocation(spec, func),
RustShape::AxumHandler => axum_invocation(spec, func),
RustShape::ClapCli => clap_invocation(spec, func),
}
}
fn actix_invocation(spec: &HarnessSpec, func: &str) -> (String, String) {
// Real actix_web requires an async runtime; the test fixtures use a
// synchronous shim signature `pub fn <func>(payload: &str) -> String`
// to keep build deps zero. The harness driver invokes it directly.
match &spec.payload_slot {
PayloadSlot::Param(0) => (String::new(), format!("let _ = entry::{func}(&payload);")),
PayloadSlot::EnvVar(name) => (
format!(" std::env::set_var({name:?}, &payload);\n"),
format!("let _ = entry::{func}(\"\");"),
),
PayloadSlot::HttpBody => (
String::new(),
format!("let _ = entry::{func}(&payload);"),
),
PayloadSlot::QueryParam(name) => (
String::new(),
format!(
"let _ = entry::{func}(&format!(\"{name}={{}}\", payload));",
),
),
_ => (String::new(), format!("let _ = entry::{func}(&payload);")),
}
}
fn axum_invocation(spec: &HarnessSpec, func: &str) -> (String, String) {
actix_invocation(spec, func)
}
fn clap_invocation(spec: &HarnessSpec, func: &str) -> (String, String) {
// Emulate clap's args by passing the payload as the sole positional
// argument. Fixture entry signature: `pub fn <func>(args: Vec<String>)`.
let pad = match &spec.payload_slot {
PayloadSlot::Argv(n) => *n,
_ => 0,
};
let mut pre = String::from(" let mut argv = vec![\"nyx_harness\".to_string()];\n");
for _ in 0..pad {
pre.push_str(" argv.push(String::new());\n");
}
pre.push_str(" argv.push(payload.clone());\n");
let call = format!("entry::{func}(argv);");
(pre, call)
}
#[cfg(test)]
mod tests {
use super::*;
@ -535,9 +692,86 @@ mod tests {
#[test]
fn entry_kind_hint_names_attempted_and_phase() {
let hint = RustEmitter.entry_kind_hint(EntryKind::HttpRoute);
assert!(hint.contains("HttpRoute"));
assert!(hint.contains("phase 16"));
let hint = RustEmitter.entry_kind_hint(EntryKind::LibraryApi);
assert!(hint.contains("LibraryApi"));
assert!(hint.contains("Phase 16"));
}
// ── Phase 16: shape detection ────────────────────────────────────────────
fn make_spec_with(kind: EntryKind, name: &str, entry_file: &str) -> HarnessSpec {
let mut s = make_spec(PayloadSlot::Param(0));
s.entry_kind = kind;
s.entry_name = name.to_owned();
s.entry_file = entry_file.to_owned();
s
}
#[test]
fn shape_detect_axum_handler() {
let src = "use axum::extract::Query; pub fn handler(payload: &str) -> String { String::new() }";
let spec = make_spec_with(EntryKind::HttpRoute, "handler", "src/entry.rs");
assert_eq!(RustShape::detect(&spec, src), RustShape::AxumHandler);
}
#[test]
fn shape_detect_actix_route() {
let src = "use actix_web::HttpResponse; pub fn handler(payload: &str) -> String { String::new() }";
let spec = make_spec_with(EntryKind::HttpRoute, "handler", "src/entry.rs");
assert_eq!(RustShape::detect(&spec, src), RustShape::ActixWebRoute);
}
#[test]
fn shape_detect_clap_cli() {
let src = "use clap::Parser; pub fn run(args: Vec<String>) {}";
let spec = make_spec_with(EntryKind::CliSubcommand, "run", "src/entry.rs");
assert_eq!(RustShape::detect(&spec, src), RustShape::ClapCli);
}
#[test]
fn shape_detect_libfuzzer_target() {
let src = "pub fn fuzz_target(data: &[u8]) {}";
let spec = make_spec_with(EntryKind::LibraryApi, "fuzz_target", "src/entry.rs");
assert_eq!(RustShape::detect(&spec, src), RustShape::LibfuzzerTarget);
}
#[test]
fn shape_detect_generic_fallback() {
let src = "pub fn run(payload: &str) {}";
let spec = make_spec_with(EntryKind::Function, "run", "src/entry.rs");
assert_eq!(RustShape::detect(&spec, src), RustShape::Generic);
}
#[test]
fn axum_shape_emits_str_invocation() {
let mut spec = make_spec_with(EntryKind::HttpRoute, "handler", "src/entry.rs");
spec.payload_slot = PayloadSlot::QueryParam("q".into());
let src = generate_main_rs(&spec, RustShape::AxumHandler);
assert!(src.contains("entry::handler"));
assert!(src.contains("q={}"));
}
#[test]
fn axum_shape_param0_passes_raw_payload() {
let spec = make_spec_with(EntryKind::HttpRoute, "handler", "src/entry.rs");
let src = generate_main_rs(&spec, RustShape::AxumHandler);
assert!(src.contains("entry::handler(&payload)"));
}
#[test]
fn clap_shape_emits_argv() {
let mut spec = make_spec_with(EntryKind::CliSubcommand, "run", "src/entry.rs");
spec.payload_slot = PayloadSlot::Argv(0);
let src = generate_main_rs(&spec, RustShape::ClapCli);
assert!(src.contains("argv.push(payload.clone())"));
assert!(src.contains("entry::run(argv)"));
}
#[test]
fn libfuzzer_shape_emits_bytes_invocation() {
let spec = make_spec_with(EntryKind::LibraryApi, "fuzz_target", "src/entry.rs");
let src = generate_main_rs(&spec, RustShape::LibfuzzerTarget);
assert!(src.contains("entry::fuzz_target(payload.as_bytes())"));
}
#[test]

View file

@ -220,6 +220,46 @@ pub fn run_spec(spec: &HarnessSpec, opts: &SandboxOptions) -> Result<RunOutcome,
_ => {}
}
}
Lang::C => {
// Compile the harness binary with `cc -o nyx_harness main.c`.
match build_sandbox::prepare_c(spec, &harness.workdir) {
Ok(build_result) => {
let binary = build_result.venv_path.join("nyx_harness");
if binary.exists() {
harness.command = vec![binary.to_string_lossy().into_owned()];
} else {
let fallback = harness.workdir.join("nyx_harness");
if fallback.exists() {
harness.command = vec![fallback.to_string_lossy().into_owned()];
}
}
}
Err(build_sandbox::BuildError::BuildFailed { stderr, attempts }) => {
return Err(RunError::BuildFailed { stderr, attempts });
}
Err(_) => {}
}
}
Lang::Cpp => {
// Compile the harness binary with `c++ -o nyx_harness main.cpp`.
match build_sandbox::prepare_cpp(spec, &harness.workdir) {
Ok(build_result) => {
let binary = build_result.venv_path.join("nyx_harness");
if binary.exists() {
harness.command = vec![binary.to_string_lossy().into_owned()];
} else {
let fallback = harness.workdir.join("nyx_harness");
if fallback.exists() {
harness.command = vec![fallback.to_string_lossy().into_owned()];
}
}
}
Err(build_sandbox::BuildError::BuildFailed { stderr, attempts }) => {
return Err(RunError::BuildFailed { stderr, attempts });
}
Err(_) => {}
}
}
_ => {
// No build step for other languages.
}

157
tests/c_fixtures.rs Normal file
View file

@ -0,0 +1,157 @@
//! C fixture integration tests (Phase 16 acceptance gate).
//!
//! Runs the dynamic verification pipeline against each C shape fixture and
//! asserts the expected verdict. Requires `--features dynamic` and `cc` on
//! PATH (override via `NYX_CC_BIN`).
//!
//! File layout per shape:
//! ```text
//! tests/dynamic_fixtures/c/<shape>/{vuln,benign}.c
//! ```
//!
//! Run with: `cargo nextest run --features dynamic --test c_fixtures`
mod common;
#[cfg(feature = "dynamic")]
mod c_fixture_tests {
use crate::common::fixture_harness::run_shape_fixture_lang;
use nyx_scanner::dynamic::spec::PayloadSlot;
use nyx_scanner::evidence::{EntryKind, VerifyResult, VerifyStatus};
use nyx_scanner::labels::Cap;
use nyx_scanner::symbol::Lang;
fn cc_available() -> bool {
let bin = std::env::var("NYX_CC_BIN").unwrap_or_else(|_| "cc".to_owned());
std::process::Command::new(&bin)
.arg("--version")
.output()
.map(|o| o.status.success())
.unwrap_or(false)
}
fn assert_confirmed(shape: &str, result: &VerifyResult) {
assert_eq!(
result.status,
VerifyStatus::Confirmed,
"{shape}/vuln: expected Confirmed, got {:?} ({:?})",
result.status,
result.detail,
);
}
fn assert_not_confirmed(shape: &str, result: &VerifyResult) {
assert!(
matches!(
result.status,
VerifyStatus::NotConfirmed | VerifyStatus::Inconclusive
),
"{shape}/benign: expected NotConfirmed (or Inconclusive), got {:?} ({:?})",
result.status,
result.detail,
);
assert_ne!(
result.status,
VerifyStatus::Confirmed,
"{shape}/benign: must not confirm",
);
}
fn run(
shape: &str,
file: &str,
func: &str,
cap: Cap,
sink_line: u32,
kind: EntryKind,
slot: PayloadSlot,
) -> VerifyResult {
run_shape_fixture_lang(
Lang::C, "c", shape, file, func, cap, sink_line, kind, slot,
)
}
// ── main_argv ───────────────────────────────────────────────────────────
#[test]
fn main_argv_vuln_is_confirmed() {
if !cc_available() {
eprintln!("SKIP: cc not available");
return;
}
let r = run(
"main_argv", "vuln.c", "nyx_entry_main", Cap::CODE_EXEC, 23,
EntryKind::CliSubcommand, PayloadSlot::Argv(0),
);
assert_confirmed("main_argv", &r);
}
#[test]
fn main_argv_benign_not_confirmed() {
if !cc_available() {
eprintln!("SKIP: cc not available");
return;
}
let r = run(
"main_argv", "benign.c", "nyx_entry_main", Cap::CODE_EXEC, 11,
EntryKind::CliSubcommand, PayloadSlot::Argv(0),
);
assert_not_confirmed("main_argv", &r);
}
// ── libfuzzer ───────────────────────────────────────────────────────────
#[test]
fn libfuzzer_vuln_is_confirmed() {
if !cc_available() {
eprintln!("SKIP: cc not available");
return;
}
let r = run(
"libfuzzer", "vuln.c", "LLVMFuzzerTestOneInput", Cap::CODE_EXEC, 16,
EntryKind::LibraryApi, PayloadSlot::Param(0),
);
assert_confirmed("libfuzzer", &r);
}
#[test]
fn libfuzzer_benign_not_confirmed() {
if !cc_available() {
eprintln!("SKIP: cc not available");
return;
}
let r = run(
"libfuzzer", "benign.c", "LLVMFuzzerTestOneInput", Cap::CODE_EXEC, 10,
EntryKind::LibraryApi, PayloadSlot::Param(0),
);
assert_not_confirmed("libfuzzer", &r);
}
// ── free_fn ─────────────────────────────────────────────────────────────
#[test]
fn free_fn_vuln_is_confirmed() {
if !cc_available() {
eprintln!("SKIP: cc not available");
return;
}
let r = run(
"free_fn", "vuln.c", "run", Cap::CODE_EXEC, 15,
EntryKind::Function, PayloadSlot::Param(0),
);
assert_confirmed("free_fn", &r);
}
#[test]
fn free_fn_benign_not_confirmed() {
if !cc_available() {
eprintln!("SKIP: cc not available");
return;
}
let r = run(
"free_fn", "benign.c", "run", Cap::CODE_EXEC, 10,
EntryKind::Function, PayloadSlot::Param(0),
);
assert_not_confirmed("free_fn", &r);
}
}

157
tests/cpp_fixtures.rs Normal file
View file

@ -0,0 +1,157 @@
//! C++ fixture integration tests (Phase 16 acceptance gate).
//!
//! Runs the dynamic verification pipeline against each C++ shape fixture
//! and asserts the expected verdict. Requires `--features dynamic` and
//! `c++` on PATH (override via `NYX_CXX_BIN`).
//!
//! File layout per shape:
//! ```text
//! tests/dynamic_fixtures/cpp/<shape>/{vuln,benign}.cpp
//! ```
//!
//! Run with: `cargo nextest run --features dynamic --test cpp_fixtures`
mod common;
#[cfg(feature = "dynamic")]
mod cpp_fixture_tests {
use crate::common::fixture_harness::run_shape_fixture_lang;
use nyx_scanner::dynamic::spec::PayloadSlot;
use nyx_scanner::evidence::{EntryKind, VerifyResult, VerifyStatus};
use nyx_scanner::labels::Cap;
use nyx_scanner::symbol::Lang;
fn cxx_available() -> bool {
let bin = std::env::var("NYX_CXX_BIN").unwrap_or_else(|_| "c++".to_owned());
std::process::Command::new(&bin)
.arg("--version")
.output()
.map(|o| o.status.success())
.unwrap_or(false)
}
fn assert_confirmed(shape: &str, result: &VerifyResult) {
assert_eq!(
result.status,
VerifyStatus::Confirmed,
"{shape}/vuln: expected Confirmed, got {:?} ({:?})",
result.status,
result.detail,
);
}
fn assert_not_confirmed(shape: &str, result: &VerifyResult) {
assert!(
matches!(
result.status,
VerifyStatus::NotConfirmed | VerifyStatus::Inconclusive
),
"{shape}/benign: expected NotConfirmed (or Inconclusive), got {:?} ({:?})",
result.status,
result.detail,
);
assert_ne!(
result.status,
VerifyStatus::Confirmed,
"{shape}/benign: must not confirm",
);
}
fn run(
shape: &str,
file: &str,
func: &str,
cap: Cap,
sink_line: u32,
kind: EntryKind,
slot: PayloadSlot,
) -> VerifyResult {
run_shape_fixture_lang(
Lang::Cpp, "cpp", shape, file, func, cap, sink_line, kind, slot,
)
}
// ── main_argv ───────────────────────────────────────────────────────────
#[test]
fn main_argv_vuln_is_confirmed() {
if !cxx_available() {
eprintln!("SKIP: c++ not available");
return;
}
let r = run(
"main_argv", "vuln.cpp", "nyx_entry_main", Cap::CODE_EXEC, 16,
EntryKind::CliSubcommand, PayloadSlot::Argv(0),
);
assert_confirmed("main_argv", &r);
}
#[test]
fn main_argv_benign_not_confirmed() {
if !cxx_available() {
eprintln!("SKIP: c++ not available");
return;
}
let r = run(
"main_argv", "benign.cpp", "nyx_entry_main", Cap::CODE_EXEC, 11,
EntryKind::CliSubcommand, PayloadSlot::Argv(0),
);
assert_not_confirmed("main_argv", &r);
}
// ── libfuzzer ───────────────────────────────────────────────────────────
#[test]
fn libfuzzer_vuln_is_confirmed() {
if !cxx_available() {
eprintln!("SKIP: c++ not available");
return;
}
let r = run(
"libfuzzer", "vuln.cpp", "LLVMFuzzerTestOneInput", Cap::CODE_EXEC, 15,
EntryKind::LibraryApi, PayloadSlot::Param(0),
);
assert_confirmed("libfuzzer", &r);
}
#[test]
fn libfuzzer_benign_not_confirmed() {
if !cxx_available() {
eprintln!("SKIP: c++ not available");
return;
}
let r = run(
"libfuzzer", "benign.cpp", "LLVMFuzzerTestOneInput", Cap::CODE_EXEC, 10,
EntryKind::LibraryApi, PayloadSlot::Param(0),
);
assert_not_confirmed("libfuzzer", &r);
}
// ── free_fn ─────────────────────────────────────────────────────────────
#[test]
fn free_fn_vuln_is_confirmed() {
if !cxx_available() {
eprintln!("SKIP: c++ not available");
return;
}
let r = run(
"free_fn", "vuln.cpp", "run", Cap::CODE_EXEC, 12,
EntryKind::Function, PayloadSlot::Param(0),
);
assert_confirmed("free_fn", &r);
}
#[test]
fn free_fn_benign_not_confirmed() {
if !cxx_available() {
eprintln!("SKIP: c++ not available");
return;
}
let r = run(
"free_fn", "benign.cpp", "run", Cap::CODE_EXEC, 10,
EntryKind::Function, PayloadSlot::Param(0),
);
assert_not_confirmed("free_fn", &r);
}
}

View file

@ -0,0 +1,11 @@
/* Phase 16 — free function with (const char *, size_t), benign. */
#include <stddef.h>
#include <stdio.h>
#include <stdlib.h>
void run(const char *payload, size_t len) {
(void)payload; (void)len;
printf("__NYX_SINK_HIT__\n");
fflush(stdout);
system("echo hello");
}

View file

@ -0,0 +1,17 @@
/* Phase 16 — free function with (const char *, size_t), vulnerable.
*
* Cap: CODE_EXEC. Concatenates payload into a shell command.
*/
#include <stddef.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
void run(const char *payload, size_t len) {
printf("__NYX_SINK_HIT__\n");
fflush(stdout);
if (!payload || len > 2048) return;
char cmd[4096];
snprintf(cmd, sizeof(cmd), "echo hello %s", payload);
system(cmd);
}

View file

@ -0,0 +1,13 @@
/* Phase 16 — libFuzzer entry, benign. */
#include <stddef.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
(void)data; (void)size;
printf("__NYX_SINK_HIT__\n");
fflush(stdout);
system("echo hello");
return 0;
}

View file

@ -0,0 +1,20 @@
/* Phase 16 — libFuzzer entry, vulnerable.
*
* Real libFuzzer entry: `int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size)`.
* Cap: CODE_EXEC.
*/
#include <stddef.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
printf("__NYX_SINK_HIT__\n");
fflush(stdout);
if (size == 0 || size > 2048) return 0;
char cmd[4096];
snprintf(cmd, sizeof(cmd), "echo hello %.*s", (int)size, (const char*)data);
system(cmd);
return 0;
}

View file

@ -0,0 +1,15 @@
/* Phase 16 — main(argc, argv), benign.
*
* Shape marker: int main(int argc, char *argv[])
* Echoes a fixed greeting; argv is ignored.
*/
#include <stdio.h>
#include <stdlib.h>
int nyx_entry_main(int argc, char *argv[]) {
(void)argc; (void)argv;
printf("__NYX_SINK_HIT__\n");
fflush(stdout);
system("echo hello");
return 0;
}

View file

@ -0,0 +1,25 @@
/* Phase 16 — main(argc, argv), vulnerable.
*
* Entry: nyx_entry_main(int argc, char *argv[])
*
* Renamed away from `main` so the harness `main` symbol does not collide
* when the entry source is `#include`d. The harness emitter recognises the
* shape via the `int main(int argc, char *argv[])` substring in the
* comment header below, then calls `nyx_entry_main` with payload-bearing
* argv. Cap: CODE_EXEC.
*
* Shape marker: int main(int argc, char *argv[])
*/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
int nyx_entry_main(int argc, char *argv[]) {
printf("__NYX_SINK_HIT__\n");
fflush(stdout);
if (argc < 2) return 0;
char cmd[4096];
snprintf(cmd, sizeof(cmd), "echo hello %s", argv[argc - 1]);
system(cmd);
return 0;
}

View file

@ -0,0 +1,12 @@
// Phase 16 — free function with (const char *, size_t), benign.
#include <cstddef>
#include <cstdio>
#include <cstdlib>
void run(const char *payload, std::size_t len) {
(void)payload; (void)len;
std::printf("__NYX_SINK_HIT__\n");
std::fflush(stdout);
std::system("echo hello");
}

View file

@ -0,0 +1,15 @@
// Phase 16 — free function with (const char *, size_t), vulnerable.
// Cap: CODE_EXEC.
#include <cstddef>
#include <cstdio>
#include <cstdlib>
#include <string>
void run(const char *payload, std::size_t len) {
std::printf("__NYX_SINK_HIT__\n");
std::fflush(stdout);
if (!payload || len > 2048) return;
std::string cmd = std::string("echo hello ") + payload;
std::system(cmd.c_str());
}

View file

@ -0,0 +1,14 @@
// Phase 16 — libFuzzer entry, benign.
#include <cstddef>
#include <cstdint>
#include <cstdio>
#include <cstdlib>
extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
(void)data; (void)size;
std::printf("__NYX_SINK_HIT__\n");
std::fflush(stdout);
std::system("echo hello");
return 0;
}

View file

@ -0,0 +1,17 @@
// Phase 16 — libFuzzer entry, vulnerable. Cap: CODE_EXEC.
#include <cstddef>
#include <cstdint>
#include <cstdio>
#include <cstdlib>
#include <string>
extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
std::printf("__NYX_SINK_HIT__\n");
std::fflush(stdout);
if (size == 0 || size > 2048) return 0;
std::string payload(reinterpret_cast<const char*>(data), size);
std::string cmd = std::string("echo hello ") + payload;
std::system(cmd.c_str());
return 0;
}

View file

@ -0,0 +1,13 @@
// Phase 16 — main(argc, argv), benign.
// Shape marker: int main(int argc, char *argv[])
#include <cstdio>
#include <cstdlib>
int nyx_entry_main(int argc, char *argv[]) {
(void)argc; (void)argv;
std::printf("__NYX_SINK_HIT__\n");
std::fflush(stdout);
std::system("echo hello");
return 0;
}

View file

@ -0,0 +1,18 @@
// Phase 16 — main(argc, argv), vulnerable.
//
// Renamed away from `main` so the harness `main` symbol does not collide.
// Shape marker: int main(int argc, char *argv[])
// Cap: CODE_EXEC.
#include <cstdio>
#include <cstdlib>
#include <string>
int nyx_entry_main(int argc, char *argv[]) {
std::printf("__NYX_SINK_HIT__\n");
std::fflush(stdout);
if (argc < 2) return 0;
std::string cmd = std::string("echo hello ") + argv[argc - 1];
std::system(cmd.c_str());
return 0;
}

View file

@ -0,0 +1,16 @@
//! Phase 16 — actix_web route, benign.
//!
//! Marker comment for shape detection: `use actix_web::HttpResponse;`
//! Echoes a fixed greeting; payload is dropped on the floor.
use std::process::Command;
pub fn handler(_payload: &str) -> String {
println!("__NYX_SINK_HIT__");
let _ = std::io::Write::flush(&mut std::io::stdout());
let out = Command::new("echo").arg("hello").output();
if let Ok(o) = out {
print!("{}", String::from_utf8_lossy(&o.stdout));
}
String::new()
}

View file

@ -0,0 +1,21 @@
//! Phase 16 — actix_web route, vulnerable.
//!
//! Marker comment for shape detection: `use actix_web::HttpResponse;`
//! The fixture exposes a synchronous shim with the same conceptual entry
//! signature so the harness build does not need to link real actix_web.
//! Cap: CODE_EXEC
use std::process::Command;
pub fn handler(payload: &str) -> String {
println!("__NYX_SINK_HIT__");
let _ = std::io::Write::flush(&mut std::io::stdout());
let out = Command::new("sh")
.arg("-c")
.arg(format!("echo hello {}", payload))
.output();
if let Ok(o) = out {
print!("{}", String::from_utf8_lossy(&o.stdout));
}
String::new()
}

View file

@ -0,0 +1,15 @@
//! Phase 16 — axum handler, benign.
//!
//! Marker comment for shape detection: `use axum::extract::Query;`
use std::process::Command;
pub fn handler(_payload: &str) -> String {
println!("__NYX_SINK_HIT__");
let _ = std::io::Write::flush(&mut std::io::stdout());
let out = Command::new("echo").arg("hello").output();
if let Ok(o) = out {
print!("{}", String::from_utf8_lossy(&o.stdout));
}
String::new()
}

View file

@ -0,0 +1,19 @@
//! Phase 16 — axum handler, vulnerable.
//!
//! Marker comment for shape detection: `use axum::extract::Query;`
//! Cap: CODE_EXEC
use std::process::Command;
pub fn handler(payload: &str) -> String {
println!("__NYX_SINK_HIT__");
let _ = std::io::Write::flush(&mut std::io::stdout());
let out = Command::new("sh")
.arg("-c")
.arg(format!("echo hello {}", payload))
.output();
if let Ok(o) = out {
print!("{}", String::from_utf8_lossy(&o.stdout));
}
String::new()
}

View file

@ -0,0 +1,14 @@
//! Phase 16 — clap-driven CLI, benign.
//!
//! Marker comment for shape detection: `use clap::Parser;`
use std::process::Command;
pub fn run(_args: Vec<String>) {
println!("__NYX_SINK_HIT__");
let _ = std::io::Write::flush(&mut std::io::stdout());
let out = Command::new("echo").arg("hello").output();
if let Ok(o) = out {
print!("{}", String::from_utf8_lossy(&o.stdout));
}
}

View file

@ -0,0 +1,20 @@
//! Phase 16 — clap-driven CLI, vulnerable.
//!
//! Marker comment for shape detection: `use clap::Parser;`
//! Signature: `pub fn run(args: Vec<String>)` — last positional arg is the
//! tainted input that is concatenated into a shell command.
use std::process::Command;
pub fn run(args: Vec<String>) {
println!("__NYX_SINK_HIT__");
let _ = std::io::Write::flush(&mut std::io::stdout());
let payload = args.last().cloned().unwrap_or_default();
let out = Command::new("sh")
.arg("-c")
.arg(format!("echo hello {}", payload))
.output();
if let Ok(o) = out {
print!("{}", String::from_utf8_lossy(&o.stdout));
}
}

View file

@ -0,0 +1,14 @@
//! Phase 16 — libfuzzer-style target, benign.
//!
//! Marker comment for shape detection: `libfuzzer_sys::fuzz_target!`
use std::process::Command;
pub fn fuzz_target(_data: &[u8]) {
println!("__NYX_SINK_HIT__");
let _ = std::io::Write::flush(&mut std::io::stdout());
let out = Command::new("echo").arg("hello").output();
if let Ok(o) = out {
print!("{}", String::from_utf8_lossy(&o.stdout));
}
}

View file

@ -0,0 +1,19 @@
//! Phase 16 — libfuzzer-style target, vulnerable.
//!
//! Marker comment for shape detection: `libfuzzer_sys::fuzz_target!`
//! Signature: `pub fn fuzz_target(data: &[u8])`.
use std::process::Command;
pub fn fuzz_target(data: &[u8]) {
println!("__NYX_SINK_HIT__");
let _ = std::io::Write::flush(&mut std::io::stdout());
let payload = String::from_utf8_lossy(data).into_owned();
let out = Command::new("sh")
.arg("-c")
.arg(format!("echo hello {}", payload))
.output();
if let Ok(o) = out {
print!("{}", String::from_utf8_lossy(&o.stdout));
}
}

View file

@ -85,9 +85,13 @@ mod verify_e2e {
}
}
/// Same as `taint_diag_with_cap` but uses a C source file so that
/// `HarnessSpec::from_finding` derives `Lang::C`, which has no emitter.
fn taint_diag_c_lang(cap: Cap) -> Diag {
/// Phase 16 turned every [`crate::symbol::Lang`] into a supported
/// emitter, so the legacy `LangUnsupported` exit path is no longer
/// reachable through `verify_finding` for any real language. The
/// helper is retained as a stub for the two tests below until they
/// are rewritten to test a different unsupported scenario.
#[allow(dead_code)]
fn taint_diag_c_lang(_cap: Cap) -> Diag {
Diag {
path: "src/handler.c".into(),
line: 10,
@ -100,14 +104,7 @@ mod verify_e2e {
message: None,
labels: vec![],
confidence: Some(Confidence::High),
evidence: Some(Evidence {
flow_steps: vec![
source_step("src/handler.c", "handle_request"),
sink_step("src/handler.c"),
],
sink_caps: cap.bits(),
..Default::default()
}),
evidence: None,
rank_score: None,
rank_reason: None,
suppressed: false,
@ -119,17 +116,17 @@ mod verify_e2e {
}
}
/// A finding with a supported cap (SQL_QUERY) and a derivable spec reaches
/// `harness::build`. The finding uses a C entry file; `Lang::C` has no
/// emitter so `LangUnsupported` is returned.
/// Phase 16 made every language emitter real, so the legacy
/// `Lang::C → LangUnsupported` exit path collapses. Retained as
/// a smoke test that an evidence-less finding still short-circuits
/// with a non-`Confirmed` verdict via `EvidenceRequired`.
#[test]
fn verify_finding_rust_lang_returns_lang_unsupported() {
fn verify_finding_without_evidence_short_circuits() {
let diag = taint_diag_c_lang(Cap::SQL_QUERY);
let opts = VerifyOptions::default();
let result = verify_finding(&diag, &opts);
assert_eq!(result.status, VerifyStatus::Unsupported);
assert_eq!(result.reason, Some(UnsupportedReason::LangUnsupported));
assert_ne!(result.status, VerifyStatus::Confirmed);
assert!(result.triggered_payload.is_none());
assert!(result.attempts.is_empty());
}
@ -161,11 +158,12 @@ mod verify_e2e {
assert_eq!(result.reason, Some(UnsupportedReason::ConfidenceTooLow));
}
/// The JSON shape of `VerifyResult` for a C finding (lang unsupported)
/// matches the documented contract: `status`, `reason` present;
/// `triggered_payload`, `detail`, `attempts` absent (skipped by serde).
/// The JSON shape of `VerifyResult` for an evidence-less finding
/// matches the documented contract: `status` present; transient
/// fields like `triggered_payload`, `detail`, `attempts` absent
/// (skipped by serde when empty / None).
#[test]
fn verify_result_json_shape_lang_unsupported() {
fn verify_result_json_shape_evidence_required() {
let diag = taint_diag_c_lang(Cap::SQL_QUERY);
let opts = VerifyOptions::default();
let result = verify_finding(&diag, &opts);
@ -173,8 +171,7 @@ mod verify_e2e {
let json = serde_json::to_string(&result).expect("VerifyResult must serialize");
let v: serde_json::Value = serde_json::from_str(&json).expect("must be valid JSON");
assert_eq!(v["status"], "Unsupported");
assert_eq!(v["reason"], "LangUnsupported");
assert!(v.get("status").is_some(), "status field must be present");
assert!(v.get("triggered_payload").is_none(), "triggered_payload must be absent");
assert!(v.get("detail").is_none(), "detail must be absent");
assert!(v.get("attempts").is_none(), "attempts must be absent (empty vec skipped)");

View file

@ -276,3 +276,175 @@ mod rust_fixture_tests {
}
}
}
// ── Phase 16: per-shape acceptance ───────────────────────────────────────────
#[cfg(feature = "dynamic")]
mod phase16_shape_tests {
use crate::common::fixture_harness::run_shape_fixture_lang;
use nyx_scanner::dynamic::spec::PayloadSlot;
use nyx_scanner::evidence::{EntryKind, VerifyResult, VerifyStatus};
use nyx_scanner::labels::Cap;
use nyx_scanner::symbol::Lang;
fn rust_available() -> bool {
std::process::Command::new("cargo")
.arg("--version")
.output()
.map(|o| o.status.success())
.unwrap_or(false)
}
fn assert_confirmed(shape: &str, result: &VerifyResult) {
assert_eq!(
result.status,
VerifyStatus::Confirmed,
"{shape}/vuln: expected Confirmed, got {:?} ({:?})",
result.status,
result.detail,
);
}
fn assert_not_confirmed(shape: &str, result: &VerifyResult) {
assert!(
matches!(
result.status,
VerifyStatus::NotConfirmed | VerifyStatus::Inconclusive
),
"{shape}/benign: expected NotConfirmed (or Inconclusive), got {:?} ({:?})",
result.status,
result.detail,
);
assert_ne!(
result.status,
VerifyStatus::Confirmed,
"{shape}/benign: must not confirm",
);
}
fn run(
shape: &str,
file: &str,
func: &str,
cap: Cap,
sink_line: u32,
kind: EntryKind,
slot: PayloadSlot,
) -> VerifyResult {
run_shape_fixture_lang(
Lang::Rust, "rust", shape, file, func, cap, sink_line, kind, slot,
)
}
// ── actix_route ─────────────────────────────────────────────────────────
#[test]
fn actix_route_vuln_is_confirmed() {
if !rust_available() {
eprintln!("SKIP: cargo not available");
return;
}
let r = run(
"actix_route", "vuln.rs", "handler", Cap::CODE_EXEC, 16,
EntryKind::HttpRoute, PayloadSlot::Param(0),
);
assert_confirmed("actix_route", &r);
}
#[test]
fn actix_route_benign_not_confirmed() {
if !rust_available() {
eprintln!("SKIP: cargo not available");
return;
}
let r = run(
"actix_route", "benign.rs", "handler", Cap::CODE_EXEC, 14,
EntryKind::HttpRoute, PayloadSlot::Param(0),
);
assert_not_confirmed("actix_route", &r);
}
// ── axum_handler ────────────────────────────────────────────────────────
#[test]
fn axum_handler_vuln_is_confirmed() {
if !rust_available() {
eprintln!("SKIP: cargo not available");
return;
}
let r = run(
"axum_handler", "vuln.rs", "handler", Cap::CODE_EXEC, 15,
EntryKind::HttpRoute, PayloadSlot::Param(0),
);
assert_confirmed("axum_handler", &r);
}
#[test]
fn axum_handler_benign_not_confirmed() {
if !rust_available() {
eprintln!("SKIP: cargo not available");
return;
}
let r = run(
"axum_handler", "benign.rs", "handler", Cap::CODE_EXEC, 13,
EntryKind::HttpRoute, PayloadSlot::Param(0),
);
assert_not_confirmed("axum_handler", &r);
}
// ── clap_cli ────────────────────────────────────────────────────────────
#[test]
fn clap_cli_vuln_is_confirmed() {
if !rust_available() {
eprintln!("SKIP: cargo not available");
return;
}
let r = run(
"clap_cli", "vuln.rs", "run", Cap::CODE_EXEC, 17,
EntryKind::CliSubcommand, PayloadSlot::Argv(0),
);
assert_confirmed("clap_cli", &r);
}
#[test]
fn clap_cli_benign_not_confirmed() {
if !rust_available() {
eprintln!("SKIP: cargo not available");
return;
}
let r = run(
"clap_cli", "benign.rs", "run", Cap::CODE_EXEC, 13,
EntryKind::CliSubcommand, PayloadSlot::Argv(0),
);
assert_not_confirmed("clap_cli", &r);
}
// ── libfuzzer_target ────────────────────────────────────────────────────
#[test]
fn libfuzzer_target_vuln_is_confirmed() {
if !rust_available() {
eprintln!("SKIP: cargo not available");
return;
}
let r = run(
"libfuzzer_target", "vuln.rs", "fuzz_target", Cap::CODE_EXEC, 15,
EntryKind::LibraryApi, PayloadSlot::Param(0),
);
assert_confirmed("libfuzzer_target", &r);
}
#[test]
fn libfuzzer_target_benign_not_confirmed() {
if !rust_available() {
eprintln!("SKIP: cargo not available");
return;
}
let r = run(
"libfuzzer_target", "benign.rs", "fuzz_target", Cap::CODE_EXEC, 13,
EntryKind::LibraryApi, PayloadSlot::Param(0),
);
assert_not_confirmed("libfuzzer_target", &r);
}
}

View file

@ -319,17 +319,17 @@ mod spec_strategies {
/// emitter's supported list surface as
/// `Inconclusive(EntryKindUnsupported { lang, attempted, supported, hint })`
/// rather than `Unsupported`. End-to-end coverage:
/// - construct an HttpRoute spec via `derive_from_callgraph_entry`
/// against a language whose emitter still advertises `[Function]`
/// only (Rust, post Phase 12 — the Python emitter now supports
/// `HttpRoute` and would short-circuit the gate);
/// - construct an HttpRoute spec against a language whose emitter
/// does not advertise `HttpRoute` (C, after Phase 16 — the C
/// emitter supports `Function`, `CliSubcommand`, `LibraryApi` but
/// not `HttpRoute`);
/// - drive it through `verify_finding`;
/// - assert the verdict shape matches the promise.
#[test]
fn entry_kind_gate_promotes_unsupported_to_inconclusive_with_hint() {
let mut diag = make_diag(
"rs.http.actix_route",
"tests/dynamic_fixtures/spec_strategies/callgraph_entry_http.rs",
"c.http.handler",
"tests/dynamic_fixtures/spec_strategies/callgraph_entry_http.c",
8,
);
let mut ev = Evidence::default();
@ -359,7 +359,7 @@ mod spec_strategies {
supported,
hint,
}) => {
assert_eq!(lang, nyx_scanner::symbol::Lang::Rust);
assert_eq!(lang, nyx_scanner::symbol::Lang::C);
assert!(matches!(attempted, EntryKind::HttpRoute));
assert!(
!supported.is_empty(),
@ -367,7 +367,7 @@ mod spec_strategies {
);
assert!(
supported.contains(&EntryKind::Function),
"Rust emitter must advertise Function support; got {supported:?}"
"C emitter must advertise Function support; got {supported:?}"
);
assert!(
!hint.is_empty(),