mirror of
https://github.com/deepseek-ai/deepseek-harness.git
synced 2026-09-08 04:01:59 +00:00
fix(code-runtime-python): cap the unknown-binding preview before JSON.stringify
The reviewer's standing item: the unknown-binding reply ran JSON.stringify on the WHOLE capped target (global + '.' + name, each up to maxValueBytes code units), allocating the escaped form — up to ~6x under control-heavy input, a multi-hundred-MB spike near the maxValueBytes ceiling that no hostile-peer bound would have admitted. The escaped preview is now built from a 1 KiB prefix of the target (enough to identify the binding); capMessage still enforces the reply budget. A forged huge-name case drives the path.
This commit is contained in:
@@ -1676,7 +1676,14 @@ export class PythonCodeRuntime extends CodeRuntime {
|
||||
// error.
|
||||
const cap = this.config.maxValueBytes
|
||||
const target = `${message.global.slice(0, cap)}.${message.name.slice(0, cap)}`
|
||||
sendReply({ type: 'reply', id: message.id, ok: false, message: capMessage(`unknown binding ${JSON.stringify(target)}`, cap) })
|
||||
// JSON.stringify on the WHOLE capped target would still allocate
|
||||
// the escaped form — up to ~6x under control-heavy input, a
|
||||
// multi-hundred-MB spike near the maxValueBytes ceiling that no
|
||||
// hostile-peer bound would have admitted. The message only needs
|
||||
// to identify the binding, so the escaped form is built from a
|
||||
// 1 KiB prefix; capMessage then enforces the reply budget.
|
||||
const preview = JSON.stringify(target.slice(0, 1024))
|
||||
sendReply({ type: 'reply', id: message.id, ok: false, message: capMessage(`unknown binding ${preview}`, cap) })
|
||||
return
|
||||
}
|
||||
void (async () => {
|
||||
|
||||
@@ -3505,6 +3505,29 @@ describe('PythonCodeRuntime — hostile peer', () => {
|
||||
expect(seenLegitCall).toBe(true)
|
||||
}, 15_000)
|
||||
|
||||
it('caps the unknown-binding preview for a huge forged name', async () => {
|
||||
// The unknown-binding reply's JSON.stringify ran on the WHOLE capped
|
||||
// target, allocating the escaped form — up to ~6x under control-heavy
|
||||
// input. The preview is now built from a 1 KiB prefix, so a forged call
|
||||
// with a huge global/name cannot spike host memory near the value ceiling;
|
||||
// the reply still identifies the binding.
|
||||
const { runtime } = await setup({ maxWallMs: 8_000, maxValueBytes: 1024 * 1024 })
|
||||
const result = await runtime.run({
|
||||
program: [
|
||||
'import os, json',
|
||||
'x = await tools.echo({"ping": True})',
|
||||
'name = "n" * 100000',
|
||||
'os.write(3, json.dumps({"type":"call","id":1,"global":"tools","name":name,"args":{}}).encode() + b"\\n")',
|
||||
'return x',
|
||||
].join('\n'),
|
||||
bindings: tools({
|
||||
echo: async args => args as CodeJsonValue,
|
||||
}),
|
||||
})
|
||||
expect(result.error).toBeUndefined()
|
||||
expect(result.value).toEqual({ ping: true })
|
||||
}, 15_000)
|
||||
|
||||
it('drops forged call frames whose ids are not the next in sequence, retaining no per-id state', async () => {
|
||||
// The host used to remember every answered id in a Set, so a program could
|
||||
// write an unbounded run of unique forged ids — each frame far below the
|
||||
|
||||
Reference in New Issue
Block a user