callgraph-verify 0.1.9

//! Local-only sweep over a user-provided real binary (e.g. a 2000s-era
//! exe built with MSVC 2005/2008 — the crate's actual domain, which cannot
//! be committed to the repo for size and licensing reasons).
//!
//! Set `CALLGRAPH_REAL_EXE` to a 32-bit PE (or DOS4GW LE) path. Skipped when
//! unset — including under `CALLGRAPH_REQUIRE_FIXTURES=1`, which governs
//! only the committed fixtures (this file can never exist in CI).
//!
//! The assertions mirror the whole-image sweeps in `c_cpp_integration.rs`:
//! decoder/byte-scan agreement and diff-cleanliness over every analyzable
//! function, plus a printed summary (run with `-- --nocapture` to see it).

use callgraph_verify::{analyze, cfg_fingerprint, diff_calls, has_direct_branch_target, Pe};

#[test]
fn real_binary_sweeps_agree_and_diff_clean() {
    let exe = match std::env::var("CALLGRAPH_REAL_EXE") {
        Ok(p) if !p.trim().is_empty() => p,
        _ => {
            eprintln!(
                "skipping: set CALLGRAPH_REAL_EXE=<path-to-32bit-exe> to sweep a real binary"
            );
            return;
        }
    };
    let pe = Pe::load(&exe).unwrap_or_else(|e| panic!("load {exe}: {e}"));
    // Only check callers that are REFERENCED (a rel32 target or a stored
    // dword): that is the crate's own definition of a plausible function
    // entry (the ORIGIN-UNREFERENCED gate). A blind extent-skip walk also
    // lands mid-gap on alignment padding, whose bytes decode to junk with
    // junk rel32 targets (e.g. a real PE at 0x497A59: a 3-byte NOP sled
    // followed by `JP`/`ADD` filler) — checking those would blame the
    // verifier for decoding non-code. The macro never analyzes such
    // addresses either: every `#[origin]` must pass the referenced gate.
    let refs = pe.referenced_addrs();
    // And only check targets INSIDE the image: a decoded `CALL` to kernel
    // space (e.g. a real PE at 0xBCAE8B `OUT DX,AL` + ... + `CALL
    // 0xDAC27BB9` — data decoded as code) is definitionally not a real
    // direct call, and the byte scan (like `referenced_addrs`) only indexes
    // in-image targets. Real missed-branch bugs always land in-image.
    let (lo, hi) = pe.image_range();

    // Walk every executable section the way the fixture sweeps do.
    let mut starts: Vec<(u64, u64)> = pe
        .sections()
        .iter()
        .filter(|s| s.is_executable())
        .map(|s| {
            let start = pe.image_base + s.virtual_address as u64;
            (start, start + s.virtual_size.max(s.raw_size) as u64)
        })
        .collect();
    starts.sort();
    assert!(!starts.is_empty(), "{exe} has no executable section");

    let mut fns = 0u32;
    let mut checked_callers = 0u32;
    let mut calls = 0u32;
    let mut disagreements = Vec::new();
    let mut unclean = Vec::new();
    for (start, end) in starts {
        let mut va = start;
        while va < end {
            if let Ok(cg) = analyze(&pe, va) {
                fns += 1;
                // Diff-cleanliness is meaningful for every analyzable entry
                // (it must never panic or contradict its own call set).
                let real: Vec<u64> = cg.call_targets().into_iter().collect();
                match diff_calls(&pe, va, &real, true) {
                    Ok(d) if d.hallucinated.is_empty() && d.missing.is_empty() => {}
                    other => unclean.push((va, format!("{other:?}"))),
                }
                // Byte-scan agreement only for referenced (plausibly real)
                // entries — see the note above.
                if refs.contains(&(va as u32)) {
                    checked_callers += 1;
                    for &t in cg.direct_calls.keys() {
                        if !(lo..hi).contains(&t) {
                            continue; // decode artifact, see note above
                        }
                        calls += 1;
                        if !has_direct_branch_target(&pe, t) {
                            disagreements.push((va, t));
                        }
                    }
                }
                va = va.saturating_add(cg.max_end.saturating_sub(va).max(1));
            } else {
                va += 1;
            }
        }
    }
    eprintln!("{exe}: swept {fns} entries ({checked_callers} referenced), {calls} direct calls");
    assert!(fns > 0, "no analyzable functions in {exe}");
    assert!(
        disagreements.is_empty(),
        "{} of {calls} CALL targets disagree with the byte scan: {disagreements:?}",
        disagreements.len()
    );
    assert!(
        unclean.is_empty(),
        "{} of {fns} functions not diff-clean: {unclean:?}",
        unclean.len()
    );
}

#[test]
fn real_le_fixup_post_images_are_referenced() {
    let exe = match std::env::var("CALLGRAPH_REAL_EXE") {
        Ok(p) if !p.trim().is_empty() => p,
        _ => {
            eprintln!(
                "skipping: set CALLGRAPH_REAL_EXE=<path-to-DOS4GW-LE> to pin fixup post images"
            );
            return;
        }
    };
    let pe = Pe::load(&exe).unwrap_or_else(|e| panic!("load {exe}: {e}"));
    if pe.fixup_count() == 0 {
        eprintln!("{exe}: no 07 fixup pairs; nothing to pin");
        return;
    }
    // Real-binary counterpart of the synthetic pe.rs test: every link-time
    // value the byte scan saw as referenced must also contribute its runtime
    // (post-fixup) image to the ORIGIN-UNREFERENCED accept set. This is the
    // `mov dword [slot],pre; call [slot]` callback pattern that motivated the
    // widening (e.g. an LE binary seeds slot 0x25AE4 with link-time 0x64ABA, so
    // post 0x74ABA must count as referenced).
    let refs = pe.referenced_addrs();
    let mut checked = 0u32;
    let mut sample: Vec<(u64, u64)> = Vec::new();
    for &(pre, post) in pe.fixup_pairs() {
        if (pre as u32) == (post as u32) || !refs.contains(&(pre as u32)) {
            continue;
        }
        assert!(
            refs.contains(&(post as u32)),
            "post-fixup image {post:#x} of referenced link-time {pre:#x} \
             missing from referenced_addrs on {exe}"
        );
        if sample.len() < 8 {
            sample.push((pre, post));
        }
        checked += 1;
        if checked >= 512 {
            break;
        }
    }
    eprintln!("{exe}: verified {checked} pre->post references; sample {sample:x?}");
    assert!(
        checked > 0,
        "{exe}: no in-image 07 pre value appeared in referenced_addrs"
    );
}

/// Real-binary regression for the 0.1.8 LE jump-table fix: on a DOS4GW LE
/// the switch table's base AND its entries are stored as link-time values
/// and are relocated by the loader, so `follow_jump_table` must translate
/// both through `Pe::fixup_post_image`. Before the fix every entry read as
/// an out-of-span pre value, `switch_cases` collapsed to 0, and a body
/// calling a hidden arm looked HALLUCINATED (the arm's CALL was never
/// harvested).
///
/// Shape-based, no address manifest: walk every *referenced* entry (a
/// plausible function start), and for each function that dispatches through
/// an indirect jump require the switch to resolve. Skipped when
/// `CALLGRAPH_REAL_EXE` is unset or the binary has no 07 fixups (PE).
#[test]
fn real_le_jump_table_entries_translate_to_post_images() {
    let exe = match std::env::var("CALLGRAPH_REAL_EXE") {
        Ok(p) if !p.trim().is_empty() => p,
        _ => {
            eprintln!(
                "skipping: set CALLGRAPH_REAL_EXE=<path-to-DOS4GW-LE> to pin jump-table translation"
            );
            return;
        }
    };
    let pe = Pe::load(&exe).unwrap_or_else(|e| panic!("load {exe}: {e}"));
    if pe.fixup_count() == 0 {
        eprintln!("{exe}: no 07 fixup pairs (not LE); jump-table translation is a no-op");
        return;
    }
    let refs = pe.referenced_addrs();
    let mut starts: Vec<(u64, u64)> = pe
        .sections()
        .iter()
        .filter(|s| s.is_executable())
        .map(|s| {
            let start = pe.image_base + s.virtual_address as u64;
            (start, start + s.virtual_size.max(s.raw_size) as u64)
        })
        .collect();
    starts.sort();

    let mut indirect_fns = 0u32;
    let mut resolved_fns = 0u32;
    let mut resolved_cases = 0u32;
    let mut sample: Vec<(u64, usize, usize)> = Vec::new();
    for (start, end) in starts {
        let mut va = start;
        while va < end {
            if refs.contains(&(va as u32)) {
                if let Ok(cg) = analyze(&pe, va) {
                    if cg.indirect_jumps > 0 {
                        indirect_fns += 1;
                        if let Ok(fp) = cfg_fingerprint(&pe, va) {
                            if fp.switch_cases >= 1 {
                                resolved_fns += 1;
                                resolved_cases += fp.switch_cases as u32;
                                if sample.len() < 8 {
                                    sample.push((va, fp.switch_cases, cg.direct_calls.len()));
                                }
                            }
                        }
                    }
                    va = va.saturating_add(cg.max_end.saturating_sub(va).max(1));
                    continue;
                }
            }
            va += 1;
        }
    }
    eprintln!(
        "{exe}: {indirect_fns} indirect-jump entries, {resolved_fns} resolved \
         ({resolved_cases} cases); sample {sample:x?}"
    );
    if indirect_fns == 0 {
        eprintln!("{exe}: no indirect-jump entries; nothing to pin");
        return;
    }
    assert!(
        resolved_fns > 0,
        "{exe}: {indirect_fns} indirect-jump entries but none resolved a switch table \
         - pre-fixup table translation regressed"
    );
}