Files
codegraph/__tests__/kernel-deep-nesting.test.ts
T
Colby MchenryandGitHub 838006c947 fix(kernel): guard the native walkers against stack overflow and defer deep files to wasm (#1581) (#1600)
Fixes #1581.

## What was wrong

`codegraph init` / `codegraph index` died with `Segmentation fault` — the whole CLI
process, not a parse worker — on a C/C++ file with very deep brace nesting (llvm's
`clang/test/Parser/parser_overflow.c`, 16,384 nested `{`). The reporter's diagnosis is
exactly right: tree-sitter's parser is iterative, so the file parses fine, and then the
native kernel's **recursive walker** (`visit_node` → `visit_for_calls_and_structure` → …,
one frame per AST level) overflowed the thread's stack. A native overflow can't be caught
the way a wasm abort can, and a parse worker is a thread of the `codegraph` process, so
the SIGSEGV took the entire indexer down — no message, no per-file fallback, no partial
index.

Two things made "just give the worker a bigger stack" the wrong fix:

- it only moves the cliff — reproduced here: the reporter's 16,384-deep file kills a
  default 4 MiB worker (rc=132 on macOS / 139 on Linux), and a 100k-deep file kills the
  8 MiB **main** thread too;
- the walkers are shared by every kernel-routed language (20 of them), and each has
  several recursion points with different frame sizes, so no single stack size is a
  provable bound.

Meanwhile the wasm path already handles this shape gracefully: its JS walker catches its
own `RangeError` per file and stores a partial result with a `parse_error`. The kernel
just needed a way to get there instead of dying.

## What this does

**The kernel guards its own recursion against the calling thread's real stack bounds and
defers a too-deep file to wasm** — the same `defer:` routing signal it already uses for
files with parse errors, which `src/extraction/kernel/index.ts` treats as "take the wasm
path for this file", silently.

- `codegraph-kernel/src/stack.rs`: per-thread stack bounds from the OS, computed once per
  thread and cached — glibc/musl `pthread_getattr_np` + `pthread_attr_getstack`, macOS
  `pthread_get_stackaddr_np` + `pthread_get_stacksize_np`, Win32
  `GetCurrentThreadStackLimits` (a hand-declared `kernel32` extern; no `windows-sys`).
  `exhausted()` is one thread-local load and one compare: true once the stack pointer is
  within a 256 KiB red zone of the limit, and it latches a flag. Where the OS can't report
  bounds it falls back to a fixed 1 MiB descent budget measured from the entry stack
  pointer — safe on anything from Node's 4 MiB worker default up. So the guard is exact on
  the 4 MiB worker, the 8 MiB main thread, and any `resourceLimits.stackSizeMb` alike.
- `stack_guard!()` (defined in `lib.rs`) is the first statement of every recursive walker
  function — all **150** self-recursive or on-cycle functions across the 15 walker modules,
  found by script (every cycle in the call graph, not just direct self-calls). It returns
  `Default::default()` (`()`, `false`, `None`, `""`) so an exhausted walk simply stops
  descending; a hook returning `false` sends its caller down the generic child walk, whose
  own guard returns at once.
- `extract_file` runs the whole walk under `stack::run_guarded`: if the flag is set
  afterwards the (truncated) result is discarded and replaced by
  `defer: nesting too deep for the native walker — wasm recovery handles it`.
- `parse-pool.ts`: a comment at `new Worker(scriptPath)` records why there is deliberately
  no `resourceLimits.stackSizeMb` bump.
- No new crates beyond `libc` as a direct unix dependency (already in `Cargo.lock`
  transitively). No wire/ABI change.

Net effect for the reporter's repo: `deep.c` goes to the wasm path, lands as
`function foo` plus a recorded parse warning, and the other 31,607 files index normally.
`CODEGRAPH_KERNEL=0` and the `exclude` workaround are no longer needed.

## Tests

**Rust unit tests** (`cargo test`, 21 passed — 7 new in `stack.rs`): the walkers for
C, C++, Rust, TypeScript and Python are driven on a **1 MiB** thread (a quarter of Node's
worker default) with 30k-deep nesting and must return `defer:` instead of crashing;
shallow files are untouched; the latch resets between runs; the OS bounds are sane on the
main thread and describe a small thread's own stack.

**`__tests__/kernel-deep-nesting.test.ts`** (new, 8 tests — skips without a staged `.node`,
fails under `CODEGRAPH_KERNEL_EXPECT=1` if the kernel is missing, like the other kernel
suites):
- every default-routed language (all 20) survives a 60k-deep expression on the main thread
  — clean result or the wasm fallback's partial result, never a crash;
- the reporter's exact 16,384-brace C file is indexed (partial) on the main thread;
- 200-deep expressions in every language still take the kernel path clean (the guard never
  trips on normal code);
- inside a **default-sized 4 MiB `worker_threads` Worker** through `dist/`: the reporter's
  `deep.c` and a 60k-deep expression in every language come back `deferred` with exit 0,
  and a normal file still extracts natively;
- end-to-end through the built CLI: `codegraph init` on a repo holding `deep.c` + `ok.c`
  exits 0 and records both files, with both functions.

**Existing kernel suites**: all 15 (`kernel-*-parity`, `kernel-scaffold`,
`kernel-retry-materialize`, `kernel-grammar-parity`) pass unchanged, 147 tests — the guard
never fires on the parity fixtures.

**Reporter's probes** (`one.js` from the issue, default 4 MiB worker, this build):
`deep.c` → `deferred`, exitCode=0 (was rc=132/139); `deep100k.c` → `deferred`, exitCode=0.
Main thread: `deep.c` / `deep100k.c` → wasm partial with
`Parse error: Maximum call stack size exceeded`; a 6,000-term binary expression and a
3,000-branch `else if` chain stay on the kernel path with clean results.

**Perf** (same `dist/`, only the `.node` swapped via `CODEGRAPH_KERNEL_PATH`; interleaved
main/new ×3, `codegraph init`, macOS arm64):

| repo | main (median) | guarded (median) | nodes / edges |
|---|---|---|---|
| express (141 files) | 0.60 s (0.58–0.65) | 0.61 s (0.58–0.61) | 1,084 / identical |
| redis (786 C/H files) | 4.44 s (4.39–4.66) | 4.49 s (4.41–4.70) | 19,942 / 76,446 identical |

Within run-to-run noise, as expected for one TLS load + compare per recursion entry.

**Linux (Docker, `node:22-bookworm`, kernel built in-container, `docker run --rm --init`)** —
the reporter's platform and the glibc `pthread_getattr_np` bounds path:

```
=== platform ===
Linux efe3cc86947b 6.12.54-linuxkit #1 SMP Tue Nov  4 21:21:47 UTC 2025 aarch64 GNU/Linux
v22.22.3
-rwxr-xr-x 1 root root 35332288 Aug 22 18:02 codegraph-kernel/prebuilds/linux-arm64/codegraph-kernel.node

=== reporter repro (issue #1581): 16,384-brace deep.c, codegraph init ===
│
└  Done

init exit code: 0
  file: deep.c
  file: deep100k.c
  file: ok.c
  function: add
  function: bar
  function: foo

=== worker probe: kernel raw extract in a default 4 MiB worker ===
deep.c: deferred
deep.c: worker exitCode=0
deep100k.c: deferred
deep100k.c: worker exitCode=0
ok.c: kernel nodes=2
ok.c: worker exitCode=0

=== cargo test stack:: (glibc pthread_getattr_np bounds path) ===
test stack::tests::os_bounds_are_sane_on_this_platform ... ok
test stack::tests::small_stack_reports_its_own_bounds ... ok
test stack::tests::normal_files_are_untouched_by_the_guard ... ok
test stack::tests::deep_braces_c_defer_instead_of_crashing ... ok
test stack::tests::latch_resets_between_runs ... ok
test stack::tests::deep_parens_cpp_rust_ts_python_defer_instead_of_crashing ... ok
test result: ok. 6 passed; 0 failed; 0 ignored; 0 measured; 15 filtered out; finished in 0.23s

=== vitest: kernel-deep-nesting + kernel-scaffold ===
✓ __tests__/kernel-scaffold.test.ts (10 tests) 30ms
✓ __tests__/kernel-deep-nesting.test.ts (8 tests) 36989ms
Test Files  2 passed (2)
Tests  18 passed (18)
```

(The pre-fix crash was reproduced on macOS — rc=132 in a default worker, rc=139 on the main thread at 100k depth — not re-run inside this container; the reporter's Linux x86_64 trace is the SIGSEGV form of the same overflow.)

**Windows (Parallels ARM64 VM, MSVC 14.44, `cargo 1.97`, kernel built on the VM,
`GetCurrentThreadStackLimits` path)**:

```
head: cbf8485 fix(kernel): guard the native walkers against stack overflow and defer deep files to wasm (#1581)
=== cargo build --release (win32-arm64) ===
    Finished `release` profile [optimized] target(s) in 2m 04s
staged: 35086848 bytes
=== cargo test (stack guard unit tests) ===
test stack::tests::normal_files_are_untouched_by_the_guard ... ok
test stack::tests::os_bounds_are_sane_on_this_platform ... ok
test stack::tests::small_stack_reports_its_own_bounds ... ok
test stack::tests::deep_braces_c_defer_instead_of_crashing ... ok
test stack::tests::latch_resets_between_runs ... ok
test stack::tests::deep_parens_cpp_rust_ts_python_defer_instead_of_crashing ... ok
test result: ok. 6 passed; 0 failed; 0 ignored; 0 measured; 15 filtered out; finished in 0.49s
=== reporter repro: codegraph init on a 16,384-brace deep.c ===
└  Done
init exit code: 0
=== vitest: deep-nesting + scaffold (CODEGRAPH_KERNEL_EXPECT=1) ===
✓ __tests__/kernel-scaffold.test.ts (10 tests) 55ms
✓ __tests__/kernel-deep-nesting.test.ts (8 tests) 67239ms
   ✓ every default-routed language survives a 60k-deep expression on the main thread 52801ms
   ✓ inside a default-sized (4 MiB) parse worker, through dist/ > defers a 60k-deep expression in every default-routed language 13050ms
   ✓ end-to-end: codegraph init on a repo holding the deep file > exits 0 and records deep.c alongside the normal files 936ms
Test Files  2 passed (2)
Tests  18 passed (18)
```

(The end-to-end test is what reads the Windows index back through `node:sqlite` — `files` = `deep.c`, `ok.c`; functions `add`, `foo`.)

Full `npm test` on this branch (macOS arm64, kernel staged): **190 files passed, 3,185 tests passed, 10 skipped, 0 failed.**

Clippy note: `cargo clippy` on the current toolchain (1.92) reports 18 pre-existing lints
(`manual_contains`, `unnecessary_to_owned`, …) in walker code this PR only touched by
inserting guard lines; none are in `stack.rs`/`lib.rs`. Left alone to keep the diff
reviewable.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

https://claude.ai/code/session_01LxZj6W6Y1SHXwvpT3uwJpK
2026-08-26 10:38:29 -05:00

282 lines
12 KiB
TypeScript

/**
* Deep-nesting safety for the native kernel (#1581).
*
* The kernel's per-language walkers recurse once per AST level. tree-sitter's
* parser is iterative, so a pathologically nested file — clang's
* `clang/test/Parser/parser_overflow.c` nests 16,384 `{`; fuzzer corpora go
* deeper — parses fine and then overflowed the WALKER's native stack. A native
* overflow is uncatchable: the parse worker is a thread of the `codegraph`
* process, so the SIGSEGV killed the whole indexer with no message, no partial
* index, no per-file fallback. Worker threads get Node's 4 MiB default stack;
* the 8 MiB main thread only moved the cliff (100k levels still died).
*
* The kernel now guards its recursion against the calling thread's real stack
* bounds (codegraph-kernel/src/stack.rs) and turns an imminent overflow into
* its `defer:` routing signal, so the file takes the wasm path — whose walker
* catches its own JS `RangeError` per file and stores a partial result with a
* `parse_error`. These tests pin that contract on every default-routed
* language, on the main thread AND inside a default-sized worker, and
* end-to-end through the built CLI.
*
* Like the other kernel suites: skipped without a staged .node; CI that
* builds the kernel sets CODEGRAPH_KERNEL_EXPECT=1 so a missing binary FAILS.
*/
import { describe, it, expect, beforeAll, beforeEach, afterEach } from 'vitest';
import * as fs from 'fs';
import * as os from 'os';
import * as path from 'path';
import { execFileSync } from 'child_process';
import { Worker } from 'worker_threads';
import { extractFromSource } from '../src/extraction';
import { initGrammars, loadGrammarsForLanguages } from '../src/extraction/grammars';
import { kernelRoutes, resetKernelForTests } from '../src/extraction/kernel';
import type { Language } from '../src/types';
const REPO = path.resolve(__dirname, '..');
const KERNEL_PATH = path.join(
REPO,
'codegraph-kernel',
'prebuilds',
`${process.platform}-${process.arch}`,
'codegraph-kernel.node'
);
const kernelBuilt = fs.existsSync(KERNEL_PATH);
const expectKernel = process.env.CODEGRAPH_KERNEL_EXPECT === '1';
const BIN = path.join(REPO, 'dist', 'bin', 'codegraph.js');
const DIST_KERNEL = path.join(REPO, 'dist', 'extraction', 'kernel');
const distBuilt = fs.existsSync(BIN) && fs.existsSync(path.join(DIST_KERNEL, 'index.js'));
/** Deep enough to overflow an 8 MiB main-thread stack on every walker. */
const PARENS_DEPTH = 60_000;
/** The reporter's exact shape: clang's parser_overflow.c nests 16,384 `{`. */
const BRACES_DEPTH = 16_384;
const CANDIDATES: Language[] = [
'typescript', 'tsx', 'javascript', 'jsx', 'java', 'python', 'go', 'c', 'cpp',
'rust', 'csharp', 'ruby', 'php', 'swift', 'kotlin', 'r', 'lua', 'luau', 'scala', 'dart',
];
const EXT: Record<string, string> = {
typescript: 'ts', tsx: 'tsx', javascript: 'js', jsx: 'jsx', java: 'java', python: 'py',
go: 'go', c: 'c', cpp: 'cpp', rust: 'rs', csharp: 'cs', ruby: 'rb', php: 'php',
swift: 'swift', kotlin: 'kt', r: 'R', lua: 'lua', luau: 'luau', scala: 'scala', dart: 'dart',
};
/** A function `f` whose body is a `depth`-deep parenthesized expression. */
function deepParens(language: Language, depth: number): string {
const open = '('.repeat(depth);
const close = ')'.repeat(depth);
switch (language) {
case 'typescript': case 'tsx': case 'javascript': case 'jsx':
return `function f() { return ${open}1${close}; }\n`;
case 'java':
return `class A {\n int f() { return ${open}1${close}; }\n}\n`;
case 'python':
return `def f():\n return ${open}1${close}\n`;
case 'go':
return `package p\n\nfunc f() int { return ${open}1${close} }\n`;
case 'c':
return `int f(void) { return ${open}1${close}; }\n`;
case 'cpp':
return `int f() { return ${open}1${close}; }\n`;
case 'rust':
return `fn f() -> i32 { ${open}1${close} }\n`;
case 'csharp':
return `class A {\n int f() { return ${open}1${close}; }\n}\n`;
case 'ruby':
return `def f\n ${open}1${close}\nend\n`;
case 'php':
return `<?php\nfunction f() { return ${open}1${close}; }\n`;
case 'swift':
return `func f() -> Int { return ${open}1${close} }\n`;
case 'kotlin':
return `fun f(): Int { return ${open}1${close} }\n`;
case 'r':
return `f <- function() {\n ${open}1${close}\n}\n`;
case 'lua': case 'luau':
return `local function f()\n return ${open}1${close}\nend\n`;
case 'scala':
return `object A {\n def f(): Int = ${open}1${close}\n}\n`;
case 'dart':
return `int f() { return ${open}1${close}; }\n`;
default:
throw new Error(`no deep fixture for ${language}`);
}
}
/** The reporter's repro: a C function body of `depth` nested blocks. */
function deepBraces(depth: number): string {
return `void foo(void) {\n${'{'.repeat(depth)}${'}'.repeat(depth)}\n}\n`;
}
const ENV_KEYS = ['CODEGRAPH_KERNEL', 'CODEGRAPH_KERNEL_LANGS', 'CODEGRAPH_KERNEL_PATH'] as const;
let savedEnv: Record<string, string | undefined>;
describe.skipIf(!kernelBuilt)('kernel deep-nesting guard (#1581)', () => {
let routed: Language[] = [];
beforeAll(async () => {
resetKernelForTests();
routed = CANDIDATES.filter((l) => kernelRoutes(l));
expect(routed.length).toBeGreaterThan(0);
await initGrammars();
await loadGrammarsForLanguages(routed);
});
beforeEach(() => {
savedEnv = Object.fromEntries(ENV_KEYS.map((k) => [k, process.env[k]]));
resetKernelForTests();
});
afterEach(() => {
for (const k of ENV_KEYS) {
if (savedEnv[k] === undefined) delete process.env[k];
else process.env[k] = savedEnv[k];
}
resetKernelForTests();
});
it('every default-routed language survives a 60k-deep expression on the main thread', () => {
const failures: string[] = [];
for (const language of routed) {
const file = `deep.${EXT[language]}`;
const source = deepParens(language, PARENS_DEPTH);
// The ONLY acceptable outcomes: a clean result (the thread's stack was
// big enough for the walk), or the wasm fallback's partial result with
// its parse_error. A native overflow would have killed this process.
const result = extractFromSource(file, source, language);
const fn = result.nodes.find((n) => n.name === 'f' && (n.kind === 'function' || n.kind === 'method'));
// R's wasm walker mints `f <- function()` only after walking the
// assignment's value, so its partial result for a file this deep holds
// just the file node — the same shape main's wasm-only path produces
// (verified with CODEGRAPH_KERNEL=0). Pre-existing and out of scope
// here; what this test pins for R is that the process survives.
if (!fn && language !== 'r') failures.push(`${language}: no function node 'f' (nodes=${result.nodes.map((n) => `${n.kind}:${n.name}`).join(',')})`);
for (const e of result.errors) {
if (!/Maximum call stack|parse_error|Parse error/.test(`${e.code} ${e.message}`)) {
failures.push(`${language}: unexpected error ${e.message}`);
}
}
}
expect(failures).toEqual([]);
}, 120_000);
it("the reporter's 16,384-brace C file is indexed (partial) instead of killing the process", () => {
const result = extractFromSource('deep.c', deepBraces(BRACES_DEPTH), 'c');
expect(result.nodes.some((n) => n.kind === 'function' && n.name === 'foo')).toBe(true);
}, 60_000);
it('shallow files still take the kernel path (the guard never trips on normal code)', () => {
// Sanity for the perf-neutral claim: a 200-deep expression is far inside
// any thread's stack, so it must come back clean with no parse_error.
for (const language of routed) {
const result = extractFromSource(`ok.${EXT[language]}`, deepParens(language, 200), language);
expect(result.errors, language).toEqual([]);
expect(result.nodes.some((n) => n.name === 'f'), language).toBe(true);
}
}, 60_000);
describe.skipIf(!distBuilt)('inside a default-sized (4 MiB) parse worker, through dist/', () => {
let tmp: string;
beforeEach(() => {
tmp = fs.mkdtempSync(path.join(os.tmpdir(), 'codegraph-deep-'));
});
afterEach(() => {
fs.rmSync(tmp, { recursive: true, force: true });
});
/**
* Run the kernel's raw extraction for `file` inside a Worker with Node's
* DEFAULT resourceLimits — exactly how ParseWorkerPool runs it. Resolves
* with the worker's exit code and what it reported; a native overflow
* would SIGSEGV/SIGILL this whole vitest process instead.
*/
function runInWorker(file: string, source: string, language: Language): Promise<{ exitCode: number; outcome: string }> {
const script = path.join(tmp, 'worker.cjs');
fs.writeFileSync(
script,
[
`const { parentPort, workerData } = require('worker_threads');`,
`const { tryKernelExtractRaw } = require(${JSON.stringify(DIST_KERNEL)});`,
`const raw = tryKernelExtractRaw(workerData.file, workerData.source, workerData.language);`,
`parentPort.postMessage(raw ? 'kernel:' + raw.counts.nodes : 'deferred');`,
].join('\n')
);
return new Promise((resolve, reject) => {
let outcome = 'no message';
const w = new Worker(script, { workerData: { file, source, language } });
w.on('message', (m: string) => { outcome = m; });
w.on('error', reject);
w.on('exit', (exitCode) => resolve({ exitCode, outcome }));
});
}
it("defers the reporter's deep.c instead of crashing the worker", async () => {
const r = await runInWorker('deep.c', deepBraces(BRACES_DEPTH), 'c');
expect(r.exitCode).toBe(0);
expect(r.outcome).toBe('deferred');
}, 60_000);
it('defers a 60k-deep expression in every default-routed language', async () => {
for (const language of routed) {
const r = await runInWorker(`deep.${EXT[language]}`, deepParens(language, PARENS_DEPTH), language);
expect(r.exitCode, language).toBe(0);
// Either the guard tripped (deferred) or the walk fit — never a crash.
expect(['deferred', 'kernel'].some((p) => r.outcome.startsWith(p)), `${language}: ${r.outcome}`).toBe(true);
}
}, 180_000);
it('still extracts a normal file natively in the worker', async () => {
const r = await runInWorker('ok.c', 'int add(int a, int b) { return a + b; }\n', 'c');
expect(r.exitCode).toBe(0);
expect(r.outcome).toMatch(/^kernel:/);
}, 30_000);
});
describe.skipIf(!distBuilt)('end-to-end: codegraph init on a repo holding the deep file', () => {
let tmp: string;
beforeEach(() => {
tmp = fs.mkdtempSync(path.join(os.tmpdir(), 'codegraph-deep-cli-'));
});
afterEach(() => {
fs.rmSync(tmp, { recursive: true, force: true });
});
it('exits 0 and records deep.c alongside the normal files', () => {
fs.writeFileSync(path.join(tmp, 'deep.c'), deepBraces(BRACES_DEPTH));
fs.writeFileSync(path.join(tmp, 'ok.c'), 'int add(int a, int b) { return a + b; }\n');
execFileSync(process.execPath, [BIN, 'init', '.'], {
cwd: tmp,
encoding: 'utf-8',
stdio: ['ignore', 'pipe', 'pipe'],
timeout: 120_000,
env: {
...process.env,
CODEGRAPH_NO_DAEMON: '1',
CODEGRAPH_WASM_RELAUNCHED: '1',
CODEGRAPH_TELEMETRY: '0',
DO_NOT_TRACK: '1',
CODEGRAPH_NO_PROMPT_HOOK: '1',
},
});
const { DatabaseSync } = require('node:sqlite') as typeof import('node:sqlite');
const db = new DatabaseSync(path.join(tmp, '.codegraph', 'codegraph.db'), { readOnly: true });
try {
const files = (db.prepare('SELECT path FROM files ORDER BY path').all() as Array<{ path: string }>).map((r) => r.path);
expect(files).toEqual(['deep.c', 'ok.c']);
const fns = (db.prepare("SELECT name FROM nodes WHERE kind = 'function' ORDER BY name").all() as Array<{ name: string }>).map((r) => r.name);
expect(fns).toEqual(['add', 'foo']);
} finally {
db.close();
}
}, 180_000);
});
});
describe.skipIf(!expectKernel)('kernel presence (CODEGRAPH_KERNEL_EXPECT=1)', () => {
it('the staged .node exists so the deep-nesting suite actually ran', () => {
expect(kernelBuilt).toBe(true);
});
});