Merge pull request #30 from niklasf/fuzz

fuzz test archive round trip and reading

.github/workflows/tests.yml

@@ -45,4 +45,5 @@ jobs:
           toolchain: ${{ matrix.rust }}
       - name: Test
         run: cargo test --verbose
-
+      - name: Compile fuzz suite
+        run: cargo check --manifest-path fuzz/Cargo.toml

fuzz/.gitignore

@@ -0,0 +1,4 @@
+target
+corpus
+artifacts
+coverage

fuzz/Cargo.toml

@@ -0,0 +1,27 @@
+[package]
+name = "ar-fuzz"
+publish = false
+edition = "2021"
+
+[package.metadata]
+cargo-fuzz = true
+
+[dependencies]
+libfuzzer-sys = "0.4"
+ar = { path = ".." }
+arbitrary = { version = "1.3.2", features = ["derive"] }
+
+[workspace]
+members = ["."] # Let fuzz suite not interfere with workspaces
+
+[[bin]]
+name = "roundtrip"
+path = "fuzz_targets/roundtrip.rs"
+
+[[bin]]
+name = "roundtrip_gnu"
+path = "fuzz_targets/roundtrip_gnu.rs"
+
+[[bin]]
+name = "read"
+path = "fuzz_targets/read.rs"

fuzz/fuzz_targets/read.rs

@@ -0,0 +1,13 @@
+#![no_main]
+
+use libfuzzer_sys::fuzz_target;
+use std::io::Read;
+
+fuzz_target!(|data: &[u8]| {
+    let mut data = data;
+    let mut archive = ar::Archive::new(&mut data);
+    while let Some(Ok(mut entry)) = archive.next_entry() {
+        let mut discard = [0; 1024];
+        let _ = entry.read(&mut discard[..]);
+    }
+});

fuzz/fuzz_targets/roundtrip.rs

@@ -0,0 +1,47 @@
+#![no_main]
+
+use ar_fuzz::{Entry, Header, Model};
+use arbitrary::{Arbitrary as _, Unstructured};
+use libfuzzer_sys::fuzz_target;
+use std::io::Read as _;
+
+fuzz_target!(|data: &[u8]| {
+    let mut u = Unstructured::new(data);
+    let model = Model::arbitrary(&mut u).expect("make arbitrary model");
+
+    if model.entries.is_empty() {
+        return; // Builder does not do anything unless there is at least one entry
+    }
+
+    let mut buffer = Vec::new();
+
+    let mut builder = ar::Builder::new(&mut buffer);
+    for entry in &model.entries {
+        if let Err(err) = builder.append(&entry.header(), &mut &entry.data[..])
+        {
+            panic!("append entry: {err} with {model:?}"); // Or just return if invalid input
+        }
+    }
+
+    let mut rountripped = Model::default();
+    let mut reader = &buffer[..];
+    let mut archive = ar::Archive::new(&mut reader);
+    while let Some(entry) = archive.next_entry() {
+        let mut entry = match entry {
+            Ok(entry) => entry,
+            Err(err) => panic!("read entry: {err} with {model:?}"),
+        };
+        rountripped.entries.push(Entry {
+            header: Header::from_ar(entry.header()),
+            data: {
+                let mut data = Vec::new();
+                if let Err(err) = entry.read_to_end(&mut data) {
+                    panic!("read entry data: {err} with {model:?}");
+                }
+                data
+            },
+        })
+    }
+
+    assert_eq!(model, rountripped);
+});

fuzz/fuzz_targets/roundtrip_gnu.rs

@@ -0,0 +1,47 @@
+#![no_main]
+
+use ar_fuzz::{Entry, Header, Model};
+use arbitrary::{Arbitrary as _, Unstructured};
+use libfuzzer_sys::fuzz_target;
+use std::io::Read as _;
+
+fuzz_target!(|data: &[u8]| {
+    let mut u = Unstructured::new(data);
+    let model = Model::arbitrary(&mut u).expect("make arbitrary model");
+
+    if model.entries.is_empty() {
+        return; // Builder does not do anything unless there is at least one entry
+    }
+
+    let mut buffer = Vec::new();
+
+    let mut builder = ar::GnuBuilder::new(&mut buffer, model.identifiers());
+    for entry in &model.entries {
+        if let Err(err) = builder.append(&entry.header(), &mut &entry.data[..])
+        {
+            panic!("append entry: {err} with {model:?}"); // Or just return if invalid input
+        }
+    }
+
+    let mut rountripped = Model::default();
+    let mut reader = &buffer[..];
+    let mut archive = ar::Archive::new(&mut reader);
+    while let Some(entry) = archive.next_entry() {
+        let mut entry = match entry {
+            Ok(entry) => entry,
+            Err(err) => panic!("read entry: {err} with {model:?}"),
+        };
+        rountripped.entries.push(Entry {
+            header: Header::from_ar(entry.header()),
+            data: {
+                let mut data = Vec::new();
+                if let Err(err) = entry.read_to_end(&mut data) {
+                    panic!("read entry data: {err} with {model:?}");
+                }
+                data
+            },
+        })
+    }
+
+    assert_eq!(model, rountripped);
+});

fuzz/src/lib.rs

@@ -0,0 +1,104 @@
+use arbitrary::{Arbitrary, Unstructured};
+
+#[derive(Clone, Default, Debug, Eq, PartialEq, Arbitrary)]
+pub struct Model {
+    pub entries: Vec<Entry>,
+}
+
+impl Model {
+    pub fn identifiers(&self) -> Vec<Vec<u8>> {
+        self.entries.iter().map(|e| e.header.identifier.0.clone()).collect()
+    }
+}
+
+#[derive(Clone, Debug, Eq, PartialEq, Arbitrary)]
+pub struct Entry {
+    pub header: Header,
+    pub data: Vec<u8>,
+}
+
+impl Entry {
+    pub fn header(&self) -> ar::Header {
+        let mut header = ar::Header::new(
+            self.header.identifier.0.clone(),
+            self.data.len() as u64,
+        );
+        header.set_mtime(self.header.mtime.0);
+        header.set_uid(self.header.uid.0);
+        header.set_gid(self.header.gid.0);
+        header.set_mode(self.header.mode.0);
+        header
+    }
+}
+
+#[derive(Clone, Debug, Eq, PartialEq, Arbitrary)]
+pub struct Header {
+    identifier: Identifier,
+    mtime: Timestamp,
+    uid: Uid,
+    gid: Uid,
+    mode: Mode,
+}
+
+impl Header {
+    pub fn from_ar(header: &ar::Header) -> Header {
+        Header {
+            identifier: Identifier(header.identifier().to_vec()),
+            mtime: Timestamp(header.mtime()),
+            uid: Uid(header.uid()),
+            gid: Uid(header.gid()),
+            mode: Mode(header.mode()),
+        }
+    }
+}
+
+#[derive(Clone, Debug, Eq, PartialEq)]
+struct Identifier(Vec<u8>);
+
+impl Arbitrary<'_> for Identifier {
+    fn arbitrary(u: &mut Unstructured<'_>) -> arbitrary::Result<Self> {
+        // TODO: Reject invalid identifiers at creation time:
+        // - Long identifiers
+        // - Invalid ASCII or invalid UTF-8?
+        // - Containing NUL
+        // - Rules for '/'?
+        // - Rules for spaces?
+        String::arbitrary(u).map(|mut v| {
+            v.retain(|ch| ch != '\0' && ch != '/' && ch != ' ');
+            while v.len() > 15 {
+                v.pop();
+            }
+            if v.is_empty() {
+                v.push('a');
+            }
+            Identifier(v.into_bytes())
+        })
+    }
+}
+
+#[derive(Copy, Clone, Debug, Eq, PartialEq)]
+struct Mode(u32);
+
+impl Arbitrary<'_> for Mode {
+    fn arbitrary(u: &mut Unstructured<'_>) -> arbitrary::Result<Self> {
+        u32::arbitrary(u).map(|v| Mode(v & 0o7777))
+    }
+}
+
+#[derive(Copy, Clone, Debug, Eq, PartialEq)]
+struct Timestamp(u64);
+
+impl Arbitrary<'_> for Timestamp {
+    fn arbitrary(u: &mut Unstructured<'_>) -> arbitrary::Result<Self> {
+        u.int_in_range(0..=999_999_999_999).map(Timestamp) // TODO: Deal with entire range
+    }
+}
+
+#[derive(Copy, Clone, Debug, Eq, PartialEq)]
+struct Uid(u32);
+
+impl Arbitrary<'_> for Uid {
+    fn arbitrary(u: &mut Unstructured<'_>) -> arbitrary::Result<Self> {
+        u.int_in_range(0..=999_999).map(Uid) // TODO: Deal with entire range (#29)
+    }
+}