classfile: aaload on 1-dim ref arrays yields element type; passes: minimizer example

crates/noxis/examples/minimizer.rs

@@ -0,0 +1,322 @@
+//! Minimizer: isolate a `VerifyError` to the smallest failing method set.
+//!
+//! The full pipeline runs on the whole jar (so hierarchy and the classpath
+//! are complete), then the transformed target is verified with the whole
+//! transformed jar behind it. Bisection deletes methods from the target
+//! only: the verifier rejects a class method-locally, so the smallest subset
+//! that still fails names the culprit.
+//!
+//! ```text
+//! cargo run --example minimizer -- \
+//!     --jar mod.jar --class a/b/C [--seed 42] [--config p.toml] [--bisect]
+//! ```
+#![allow(clippy::unwrap_used, clippy::expect_used)]
+
+use std::collections::BTreeSet;
+use std::path::{Path, PathBuf};
+use std::process::{Command, ExitCode};
+
+use noxis_classfile::format;
+use noxis_classfile::model::ClassFile;
+use noxis_core::{Config, Context, PassId, Profile};
+use noxis_jar::{Entry, load, read_file, save, write_file};
+
+fn main() -> ExitCode {
+    match run() {
+        Ok(true) => ExitCode::SUCCESS,
+        Ok(false) => ExitCode::FAILURE,
+        Err(error) => {
+            eprintln!("minimizer: {error}");
+            ExitCode::from(2)
+        }
+    }
+}
+
+struct Args {
+    jar: String,
+    class: String,
+    seed: u64,
+    config: Option<String>,
+    bisect: bool,
+}
+
+fn parse_args() -> Result<Args, String> {
+    let mut args = Args {
+        jar: String::new(),
+        class: String::new(),
+        seed: 1,
+        config: None,
+        bisect: false,
+    };
+    let mut iter = std::env::args().skip(1);
+    while let Some(flag) = iter.next() {
+        match flag.as_str() {
+            "--jar" => args.jar = iter.next().ok_or("--jar needs a value")?,
+            "--class" => args.class = iter.next().ok_or("--class needs a value")?,
+            "--seed" => {
+                args.seed = iter
+                    .next()
+                    .ok_or("--seed needs a value")?
+                    .parse()
+                    .map_err(|_| "--seed must be a number")?
+            }
+            "--config" => args.config = Some(iter.next().ok_or("--config needs a value")?),
+            "--bisect" => args.bisect = true,
+            other => return Err(format!("unknown flag {other}")),
+        }
+    }
+    if args.jar.is_empty() || args.class.is_empty() {
+        return Err(
+            "usage: --jar <jar> --class <internal/name> [--seed N] [--config p.toml] [--bisect]"
+                .into(),
+        );
+    }
+    Ok(args)
+}
+
+fn load_config(args: &Args) -> Result<Config, String> {
+    let mut config = match &args.config {
+        Some(path) => {
+            let text = std::fs::read_to_string(path).map_err(|e| format!("{path}: {e}"))?;
+            Config::from_toml(&text).map_err(|e| e.to_string())?
+        }
+        None => Config {
+            profile: Profile::Paranoid,
+            ..Config::default()
+        },
+    };
+    config.seed = Some(args.seed);
+    // Junk classes never touch existing frames and only clutter the list.
+    config.disable.push(PassId::Noise);
+    Ok(config)
+}
+
+#[derive(Clone, Copy, PartialEq, Eq)]
+enum Verdict {
+    Verified,
+    Failed,
+    Skipped,
+}
+
+/// Verify a class with the full transformed jar behind it on the classpath.
+fn verify(classpath: &Path, name: &str, bytes: &[u8]) -> (Verdict, String) {
+    let scratch = std::env::temp_dir().join(format!("noxis-min-{}", std::process::id()));
+    let _ = std::fs::remove_dir_all(&scratch);
+    let mut path = scratch.clone();
+    for part in name.split('/') {
+        path.push(part);
+    }
+    path.set_extension("class");
+    let written =
+        std::fs::create_dir_all(path.parent().unwrap()).and_then(|()| std::fs::write(&path, bytes));
+    if let Err(error) = written {
+        return (Verdict::Skipped, format!("write: {error}"));
+    }
+    // scratch first so the (subset) target shadows its full-jar entry.
+    let cp = format!("{}:{}", scratch.display(), classpath.display());
+    let output = Command::new("java")
+        .arg("-Xverify:all")
+        .arg("-cp")
+        .arg(&cp)
+        .arg(name.replace('/', "."))
+        .output();
+    let _ = std::fs::remove_dir_all(&scratch);
+    let Ok(output) = output else {
+        return (Verdict::Skipped, "java failed to start".into());
+    };
+    let stderr = String::from_utf8_lossy(&output.stderr).into_owned();
+    if stderr.contains("VerifyError") || stderr.contains("ClassFormatError") {
+        (Verdict::Failed, stderr)
+    } else if stderr.contains("NoClassDefFoundError")
+        || stderr.contains("ClassNotFoundException")
+        || stderr.contains("UnsupportedClassVersionError")
+    {
+        (Verdict::Skipped, stderr)
+    } else {
+        (Verdict::Verified, stderr)
+    }
+}
+
+/// Extract the failing class and method from a VerifyError's `Location:` line:
+/// `    pkg/Class.method(desc)R @pc: insn`. The verifier blames the exact
+/// method, which may live in a superclass of the class under test. Note the
+/// class name is printed verbatim (an obfuscated name may itself start
+/// with `L`), so both spellings are tried by the caller.
+fn failing_location(stderr: &str) -> Option<(String, String)> {
+    let mut lines = stderr.lines();
+    while let Some(line) = lines.next() {
+        if line.trim() == "Location:" {
+            let loc = lines.next()?.trim();
+            let dot = loc.find('.')?;
+            let class = loc[..dot].to_string();
+            let method = loc[dot + 1..].split(" @").next()?.to_string();
+            return Some((class, method));
+        }
+    }
+    None
+}
+
+/// Method identity as the verifier sees it: name plus descriptor. All
+/// methods participate, including `<init>`/`<clinit>`: deletion is
+/// verification-safe (no code runs), and the bug may live there.
+fn method_keys(class: &ClassFile) -> Vec<(String, String)> {
+    class
+        .methods
+        .iter()
+        .map(|m| (m.name.clone(), m.descriptor.clone()))
+        .collect()
+}
+
+fn write_subset(class: &ClassFile, keep: &BTreeSet<(String, String)>) -> Option<Vec<u8>> {
+    let mut cut = class.clone();
+    cut.methods
+        .retain(|m| keep.contains(&(m.name.clone(), m.descriptor.clone())));
+    format::write(&cut).ok()
+}
+
+fn subset_verdict(cp: &Path, class: &ClassFile, keep: &BTreeSet<(String, String)>) -> Verdict {
+    let Some(bytes) = write_subset(class, keep) else {
+        return Verdict::Skipped;
+    };
+    let (verdict, stderr) = verify(cp, &class.this_class, &bytes);
+    if verdict == Verdict::Failed {
+        let reason = stderr
+            .lines()
+            .find(|l| l.contains("VerifyError"))
+            .unwrap_or("?");
+        println!("  keep={}: {reason}", keep.len());
+    }
+    verdict
+}
+
+fn bisect(cp: &Path, class: &ClassFile) -> Option<BTreeSet<(String, String)>> {
+    let mut keep: BTreeSet<(String, String)> = method_keys(class).into_iter().collect();
+    println!("bisect: {} methods total", keep.len());
+    while keep.len() > 1 {
+        let all: Vec<(String, String)> = keep.iter().cloned().collect();
+        let half = all.len() / 2;
+        let left: BTreeSet<_> = all[..half].iter().cloned().collect();
+        let right: BTreeSet<_> = all[half..].iter().cloned().collect();
+        match (
+            subset_verdict(cp, class, &left),
+            subset_verdict(cp, class, &right),
+        ) {
+            (Verdict::Failed, _) => keep = left,
+            (_, Verdict::Failed) => keep = right,
+            _ => {
+                println!("bisect: neither half fails alone - interaction; keeping all");
+                return Some(keep);
+            }
+        }
+    }
+    Some(keep)
+}
+
+fn run() -> Result<bool, String> {
+    let args = parse_args()?;
+    let jar_path = PathBuf::from(&args.jar);
+    let config = load_config(&args)?;
+    let entries = read_file(&jar_path).map_err(|e| e.to_string())?;
+    if !entries
+        .iter()
+        .any(|e| e.name == format!("{}.class", args.class))
+    {
+        return Err(format!("{} not in jar", args.class));
+    }
+    let mut program = load(entries).map_err(|e| e.to_string())?;
+    println!("loaded {} classes", program.classes.len());
+
+    let mut ctx = Context::new(config);
+    noxis_passes::standard_pipeline()
+        .run(&mut ctx, &mut program)
+        .map_err(|e| e.to_string())?;
+
+    let new_name = ctx
+        .mapping
+        .class(&args.class)
+        .unwrap_or(&args.class)
+        .to_string();
+    println!("target {} -> {new_name}", args.class);
+
+    // Whole transformed jar becomes the verifier's classpath.
+    let out_jar = std::env::temp_dir().join(format!("noxis-min-cp-{}.jar", std::process::id()));
+    let transformed: Vec<Entry> = save(&program).map_err(|e| e.to_string())?;
+    write_file(&out_jar, &transformed).map_err(|e| e.to_string())?;
+
+    let target = program
+        .class(&new_name)
+        .ok_or_else(|| format!("{new_name} vanished"))?;
+    let bytes = format::write(target).map_err(|e| e.to_string())?;
+    let (verdict, stderr) = verify(&out_jar, &new_name, &bytes);
+    match verdict {
+        Verdict::Verified => {
+            println!("PASS: this run does not reproduce the failure");
+            let _ = std::fs::remove_file(&out_jar);
+            return Ok(false);
+        }
+        Verdict::Skipped => {
+            println!("SKIP: dependencies unresolved:\n{stderr}");
+            let _ = std::fs::remove_file(&out_jar);
+            return Ok(false);
+        }
+        Verdict::Failed => {
+            println!(
+                "reproduced:\n{}",
+                stderr.lines().take(6).collect::<Vec<_>>().join("\n")
+            );
+        }
+    }
+
+    let parsed = format::parse(&bytes).map_err(|e| e.to_string())?;
+    if !args.bisect {
+        let _ = std::fs::remove_file(&out_jar);
+        return Ok(false);
+    }
+    // The verifier blames the exact method, which may live in a superclass:
+    // pivot the bisection to the class that actually fails.
+    let mut pivot_name = new_name.clone();
+    let mut pivot = parsed;
+    if let Some((fail_class, fail_method)) = failing_location(&stderr) {
+        println!("verifier blames {fail_class}.{fail_method}");
+        if fail_class != new_name {
+            // The name may itself start with `L`; HotSpot prints it verbatim.
+            let real_bytes = if let Some(real) = program.class(&fail_class) {
+                format::write(real).map_err(|e| e.to_string())?
+            } else if let Some(stripped) =
+                fail_class.strip_prefix('L').and_then(|s| program.class(s))
+            {
+                format::write(stripped).map_err(|e| e.to_string())?
+            } else {
+                return Err(format!("{fail_class} not in program"));
+            };
+            pivot = format::parse(&real_bytes).map_err(|e| e.to_string())?;
+            pivot_name = fail_class;
+            println!("pivoted to {pivot_name}");
+        }
+    }
+    let minimal = bisect(&out_jar, &pivot);
+    let _ = std::fs::remove_file(&out_jar);
+    let Some(minimal) = minimal else {
+        return Ok(false);
+    };
+    println!("\nminimal failing set in {pivot_name}:");
+    for (name, descriptor) in &minimal {
+        let method = pivot
+            .methods
+            .iter()
+            .find(|m| &m.name == name && &m.descriptor == descriptor)
+            .unwrap();
+        let insns = method.code.as_ref().map_or(0, |c| c.insns.len());
+        let frames = method.code.as_ref().map_or(0, |c| c.frames.len());
+        println!("  {name}{descriptor}  insns={insns} frames={frames}");
+    }
+    // Write the minimal single-failing-subset class for javap inspection.
+    let minimal_set: BTreeSet<(String, String)> = minimal.iter().cloned().collect();
+    if let Some(minimal_bytes) = write_subset(&pivot, &minimal_set) {
+        let dump = std::env::temp_dir().join("noxis-minimal-fail.class");
+        if std::fs::write(&dump, &minimal_bytes).is_ok() {
+            println!("minimal class bytes: {}", dump.display());
+        }
+    }
+    Ok(false)
+}