why choose 15k as MAX_INPUT_LEN?

[zjuchenyuan]

In common/src/config.rs

Angora/common/src/config.rs

Line 41 in 1940eb5

pub const MAX_INPUT_LEN: usize = 15000;

This value is used to ignore those seed files larger than 15000 bytes.

Angora/fuzzer/src/depot/sync.rs

Lines 14 to 35 in 1940eb5

	pub fn sync_depot(executor: &mut Executor, running: Arc<AtomicBool>, dir: &Path) {
	executor.local_stats.clear();
	let seed_dir = dir.read_dir().expect("read_dir call failed");
	for entry in seed_dir {
	if let Ok(entry) = entry {
	if !running.load(Ordering::SeqCst) {
	break;
	}
	let path = &entry.path();
	if path.is_file() {
	let file_len =
	fs::metadata(path).expect("Could not fetch metadata.").len() as usize;
	if file_len < config::MAX_INPUT_LEN {
	let buf = read_from_file(path);
	executor.run_sync(&buf);
	}
	}
	}
	}
	info!("sync {} file from seeds.", executor.local_stats.num_inputs);
	executor.update_log();
	}

From Line 26, only size less than 14.64kb is executed. Larger files are silently ignored, without print any warning message.

I think 14.64kb is a rather small size, why do you choose this value?

[spinpx]

If the size of seed input is too large, Angora will take too much time in byte-level taint analysis.
We had added some warnnings now.
commit : 16cb612

I will improve this problem recently.