rllvm-query 0.9.0

//! Shared `ProgramFacts`/`Session` builders for this crate's module tests.
//!
//! Kept in one module, rather than duplicated per test module, because Rust
//! test modules cannot import each other's `#[cfg(test)] mod tests` and more
//! than one submodule's tests build on these fixtures.

use std::path::{Path, PathBuf};

use rllvm_core::{
    catalog::{
        CatalogOrigin, CatalogScope, ModuleCatalog, ModuleRecord, ModuleStatus, hash_bytes,
        write_catalog,
    },
    config::LlvmBindir,
};

use crate::{
    bind::{BindingCandidate, BindingStatus, SymbolBinding},
    facts::{
        CallSiteFact, CallSiteId, CallTarget, FieldBasis, FieldEvidence, FieldRef, FunctionFact,
        FunctionId, Language, LanguageBasis, Linkage, ModuleAnalysis, ModuleReport, ProgramFacts,
        SourceLanguage, SourceLocation, SourceStatus, UseFact, UseKind,
    },
    index::Session,
};

pub(crate) fn function(
    module: &str,
    symbol: &str,
    is_definition: bool,
    linkage: Linkage,
) -> FunctionFact {
    FunctionFact {
        id: FunctionId {
            module_id: module.into(),
            symbol: symbol.into(),
        },
        is_definition,
        linkage,
        language: None,
        signature: "i32 (i32, i32)".into(),
        location: None,
        mapped_lines: Default::default(),
        alias_of: None,
    }
}

/// A definition in module `m`, attributed to `language` by its producer.
pub(crate) fn attributed(
    symbol: &str,
    linkage: Linkage,
    language: Option<Language>,
) -> FunctionFact {
    let mut fact = function("m", symbol, true, linkage);
    fact.language = language.map(|name| SourceLanguage {
        name,
        basis: LanguageBasis::Producer,
    });
    fact
}

/// A current, uninlined `file:line`. Only the two fields a fixture ever
/// varies are parameters; the rest are the defaults every caller wanted.
pub(crate) fn source_location(file: &str, line: u32) -> SourceLocation {
    SourceLocation {
        file: file.into(),
        directory: None,
        line,
        column: 1,
        source_status: SourceStatus::Current,
        status_basis: None,
        inlined_at: Vec::new(),
    }
}

pub(crate) fn direct_call(
    caller: &FunctionFact,
    callee: &FunctionFact,
    index: u32,
) -> CallSiteFact {
    CallSiteFact {
        id: CallSiteId {
            function: caller.id.clone(),
            block_index: 0,
            instruction_index: index,
        },
        location: None,
        target: CallTarget::Direct {
            callee: callee.id.clone(),
        },
    }
}

/// An indirect call site in `caller`: `llvm_target_bound` is what CVP
/// managed to bound it to, or `None` for a site it could not bound.
pub(crate) fn indirect_call(
    caller: &FunctionFact,
    index: u32,
    location: Option<SourceLocation>,
    llvm_target_bound: Option<Vec<FunctionId>>,
) -> CallSiteFact {
    CallSiteFact {
        id: CallSiteId {
            function: caller.id.clone(),
            block_index: 0,
            instruction_index: index,
        },
        location,
        target: CallTarget::Indirect {
            signature: "i32 (i32, i32)".into(),
            llvm_target_bound,
            via_field: None,
            via_slot: None,
        },
    }
}

/// An unbounded indirect call in `caller` at `t.c:line`, dispatching through
/// `field` when the pointer was traced to one.
pub(crate) fn indirect_call_through(
    caller: &FunctionFact,
    line: u32,
    field: Option<FieldRef>,
) -> CallSiteFact {
    let mut site = indirect_call(caller, line, Some(source_location("t.c", line)), None);
    if let CallTarget::Indirect { via_field, .. } = &mut site.target {
        *via_field = field.map(|field| FieldEvidence {
            field,
            basis: FieldBasis::StructGep,
            name: None,
        });
    }
    site
}

/// A module report with only its id and status set; fixtures override the
/// one or two other fields they actually pin.
pub(crate) fn report(id: &str, status: ModuleAnalysis) -> ModuleReport {
    ModuleReport {
        id: id.into(),
        status,
        ir_stage: None,
        debug_info: None,
        compiler: None,
        configuration_id: None,
        content_sha256: None,
        target_triple: None,
        producers: Vec::new(),
        diagnostic: None,
    }
}

/// An ambiguous binding for `symbol`, declared in `declared_in` and defined
/// once per `(module, configuration)` candidate.
pub(crate) fn ambiguous_binding(
    symbol: &str,
    declared_in: &[&str],
    candidates: &[(&str, Option<&str>)],
) -> SymbolBinding {
    SymbolBinding {
        symbol: symbol.into(),
        declared_in: declared_in.iter().map(|id| (*id).to_string()).collect(),
        candidates: candidates
            .iter()
            .map(|(module, configuration)| BindingCandidate {
                function: FunctionId {
                    module_id: (*module).into(),
                    symbol: symbol.into(),
                },
                configuration_id: configuration.map(str::to_string),
            })
            .collect(),
        status: BindingStatus::Ambiguous,
    }
}

pub(crate) fn facts(functions: Vec<FunctionFact>, call_sites: Vec<CallSiteFact>) -> ProgramFacts {
    ProgramFacts {
        functions,
        call_sites,
        uses: Vec::new(),
        slot_flows: Vec::new(),
        field_flows: Vec::new(),
        returned_addresses: Vec::new(),
        scope: CatalogScope {
            kind: "test".into(),
            total_entries: 1,
            selected_entries: 1,
            selection: Default::default(),
            selection_history: Vec::new(),
            analysis_arguments: Vec::new(),
            whole_program_complete: None,
            limitations: Vec::new(),
        },
        origin: CatalogOrigin {
            kind: "test".into(),
            input: PathBuf::from("test"),
            sha256: None,
        },
        modules: Vec::new(),
    }
}

/// A catalog of module `m`: `source`, written to `scratch` as `file_name`
/// and compiled with `flags`. Opened as a session whose LLVM bindir is the
/// configured toolchain's, so it can promote the module.
pub(crate) fn compiled_session(
    scratch: &Path,
    file_name: &str,
    source: &str,
    flags: &[&str],
) -> Session {
    let source_path = scratch.join(file_name);
    std::fs::write(&source_path, source).unwrap();
    let module = rllvm_testkit::compile_bitcode_file(&source_path, flags);
    module_session(scratch, &module)
}

/// A module whose debug info is from an unsupported version, which LLVM
/// strips with a warning when it reads it. `f` passes its parameter to
/// `sink`.
const STALE_DEBUG_INFO_IR: &str = r#"
declare void @sink(i32)

define void @f(i32 %x) !dbg !5 {
  call void @sink(i32 %x), !dbg !8
  ret void, !dbg !8
}

!llvm.dbg.cu = !{!0}
!llvm.module.flags = !{!3}
!0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang", emissionKind: FullDebug)
!1 = !DIFile(filename: "t.c", directory: "/tmp")
!3 = !{i32 2, !"Debug Info Version", i32 2}
!5 = distinct !DISubprogram(name: "f", scope: !1, file: !1, line: 1, type: !9, unit: !0, spFlags: DISPFlagDefinition)
!8 = !DILocation(line: 1, column: 1, scope: !5)
!9 = !DISubroutineType(types: !{})
"#;

/// [`STALE_DEBUG_INFO_IR`] assembled into `scratch`, its debug info kept:
/// `llvm-as` would strip it as LLVM's reader does.
pub(crate) fn stale_debug_info_module(scratch: &Path) -> PathBuf {
    let source = scratch.join("stale.ll");
    std::fs::write(&source, STALE_DEBUG_INFO_IR).unwrap();
    let module = scratch.join("stale.bc");
    let status = std::process::Command::new(rllvm_testkit::llvm_bin("llvm-as"))
        .arg("-disable-auto-upgrade-debug-info")
        .arg(&source)
        .arg("-o")
        .arg(&module)
        .status()
        .unwrap();
    assert!(status.success(), "llvm-as failed on {}", source.display());
    module
}

/// A catalog of the one bitcode file `module`, which lies in `scratch`, as
/// module `m`. Opened as [`compiled_session`] opens its own.
pub(crate) fn module_session(scratch: &Path, module: &Path) -> Session {
    let mut record = ModuleRecord::new("m");
    record.path = Some(module.file_name().unwrap().into());
    record.content_sha256 = Some(hash_bytes(&std::fs::read(module).unwrap()));
    record.status = ModuleStatus::Available;
    let catalog = ModuleCatalog::new(
        CatalogOrigin {
            kind: "test".into(),
            input: PathBuf::from("test"),
            sha256: None,
        },
        "test",
        vec![record],
    );
    let path = scratch.join("catalog.json");
    write_catalog(&path, &catalog).unwrap();
    let mut session = crate::open(&path).unwrap();
    let opt = rllvm_testkit::llvm_bin("opt");
    session.set_llvm_bindir(Ok(LlvmBindir::configured(opt.parent().unwrap())));
    session
}

/// One module, one edge per pair.
pub(crate) fn session_from(edges: &[(&str, &str)]) -> Session {
    let mut functions: Vec<FunctionFact> = Vec::new();
    let lookup = |functions: &mut Vec<FunctionFact>, name: &str| -> FunctionFact {
        if let Some(existing) = functions.iter().find(|f| f.id.symbol == name) {
            return existing.clone();
        }
        let created = function("m", name, true, Linkage::Internal);
        functions.push(created.clone());
        created
    };
    let mut call_sites = Vec::new();
    for (index, (from, to)) in edges.iter().enumerate() {
        let caller = lookup(&mut functions, from);
        let callee = lookup(&mut functions, to);
        call_sites.push(direct_call(&caller, &callee, index as u32));
    }
    Session::new(facts(functions, call_sites), Vec::new())
}

/// `a` reaches `target` only through an indirect site CVP bounded to
/// `{target, other}`, recorded at `t.c:9`. A second bound member matters: it
/// lets a test tell a correct bounded-indirect edge apart from one that
/// collapsed the bound to just the target taken. The location makes the site
/// addressable by `at` and `indirect-targets`, which
/// `session_with_address_taken_function` can only answer for an *unbounded*
/// site.
pub(crate) fn session_with_bounded_indirect() -> Session {
    let a = function("m", "a", true, Linkage::Internal);
    let target = function("m", "target", true, Linkage::Internal);
    let other = function("m", "other", true, Linkage::Internal);
    let site = indirect_call(
        &a,
        0,
        Some(source_location("t.c", 9)),
        Some(vec![target.id.clone(), other.id.clone()]),
    );
    Session::new(facts(vec![a, target, other], vec![site]), Vec::new())
}

/// `a` calls `b`; `b` reaches `c` only through an indirect call CVP could not
/// bound, so no path exists over direct edges.
pub(crate) fn session_with_indirect_gap() -> Session {
    let a = function("m", "a", true, Linkage::Internal);
    let b = function("m", "b", true, Linkage::Internal);
    let c = function("m", "c", true, Linkage::Internal);
    let call = direct_call(&a, &b, 0);
    let indirect = indirect_call(&b, 1, None, None);
    Session::new(facts(vec![a, b, c], vec![call, indirect]), Vec::new())
}

/// One `i32 (i32, i32)` indirect site at `t.c:4`, an `add` of that signature
/// whose address is taken, and a `log` whose address is taken but whose
/// signature does not match.
pub(crate) fn session_with_address_taken_function() -> Session {
    let add = function("m", "add", true, Linkage::Internal);
    let log = FunctionFact {
        signature: "void (ptr)".into(),
        ..function("m", "log", true, Linkage::Internal)
    };
    let caller = function("m", "caller", true, Linkage::Internal);
    let mut base = facts(
        vec![add.clone(), log.clone(), caller.clone()],
        vec![indirect_call(
            &caller,
            0,
            Some(SourceLocation {
                file: "t.c".into(),
                directory: None,
                line: 4,
                column: 3,
                source_status: SourceStatus::Unknown,
                status_basis: None,
                inlined_at: Vec::new(),
            }),
            None,
        )],
    );
    base.uses = [add.id, log.id]
        .into_iter()
        .map(|used| UseFact {
            used,
            in_function: Some(caller.id.clone()),
            in_global: None,
            location: None,
            kind: UseKind::StoredToMemory,
            into_slot: None,
            field: None,
        })
        .collect();
    Session::new(base, Vec::new())
}

/// `scope.selected_entries == 2`, but only one module was read: `a` parsed,
/// `b` failed. Pins that `scope` is quoted from the catalog and never
/// recomputed from what `analysis` actually managed to read.
pub(crate) fn facts_with_one_failed_module() -> ProgramFacts {
    let mut base = facts(vec![], vec![]);
    base.scope.total_entries = 2;
    base.scope.selected_entries = 2;
    base.modules = vec![
        ModuleReport {
            debug_info: Some(true),
            ..report("a", ModuleAnalysis::Analyzed)
        },
        ModuleReport {
            diagnostic: Some("truncated".into()),
            ..report("b", ModuleAnalysis::Failed)
        },
    ];
    base
}

/// A function whose only mapped line is the given `(file, line)` pairs.
pub(crate) fn session_from_source_lines(lines: &[(&str, u32)]) -> Session {
    let mut only = function("m", "only", true, Linkage::Internal);
    only.mapped_lines = lines
        .iter()
        .map(|(file, line)| (PathBuf::from(file), *line))
        .collect();
    Session::new(facts(vec![only], Vec::new()), Vec::new())
}

/// Two ambiguous symbols, and one of them reachable through two different
/// declarations: `c1` and `c2` each define `caller` and declare `target`,
/// while `a` and `b` each define `target`. A second binding, `unused`, is
/// ambiguous too but sits on no path out of `caller`.
///
/// Distinguishes three counts that are easy to conflate: the bindings one
/// walk reached (one), the declarations it reached them through (two), and
/// every ambiguous binding in scope (two).
pub(crate) fn session_with_ambiguous_bindings() -> Session {
    let mut functions = Vec::new();
    let mut call_sites = Vec::new();
    for (index, module) in ["c1", "c2"].into_iter().enumerate() {
        let caller = function(module, "caller", true, Linkage::External);
        let declaration = function(module, "target", false, Linkage::External);
        call_sites.push(direct_call(&caller, &declaration, index as u32));
        functions.push(caller);
        functions.push(declaration);
    }
    functions.push(function("a", "target", true, Linkage::External));
    functions.push(function("b", "target", true, Linkage::External));

    let target = ambiguous_binding(
        "target",
        &["c1", "c2"],
        &[("a", Some("debug")), ("b", Some("release"))],
    );
    let unused = ambiguous_binding("unused", &["z"], &[("a", None), ("b", None)]);
    Session::new(facts(functions, call_sites), vec![target, unused])
}

/// A C++ module: two instantiations of `twice`, a `twice_helper` that must
/// not be confused for one, `ns::twice` in a namespace, and a C `main`.
///
/// Symbols are the real mangled spellings, so `Session::new` demangles them
/// through the same path a captured module would.
pub(crate) fn session_with_cxx_symbols() -> Session {
    let functions = [
        "_Z5twiceIiET_S0_",  // int twice<int>(int)
        "_Z5twiceIdET_S0_",  // double twice<double>(double)
        "_Z12twice_helperv", // twice_helper()
        "_ZN2ns5twiceEv",    // ns::twice()
        "main",
    ]
    .into_iter()
    .map(|symbol| function("m", symbol, true, Linkage::External))
    .collect();
    Session::new(facts(functions, Vec::new()), Vec::new())
}

/// Two hashed, analyzed modules for the overlay. `mod_a` holds an unresolved
/// site through `ops@8` in `dispatch` (`t.c:4`), an unresolved site with no
/// field in `call_plain` (`t.c:5`), a site CVP bounded to `{h1}` in
/// `bounded` (`t.c:6`), a direct call from `init` to `h3`, and the handlers
/// `h1` and `h3`. Both modules define a static `helper`.
pub(crate) fn facts_with_overlay_sites() -> ProgramFacts {
    let dispatch = function("mod_a", "dispatch", true, Linkage::External);
    let call_plain = function("mod_a", "call_plain", true, Linkage::External);
    let bounded = function("mod_a", "bounded", true, Linkage::External);
    let init = function("mod_a", "init", true, Linkage::External);
    let h1 = function("mod_a", "h1", true, Linkage::Internal);
    let h3 = function("mod_a", "h3", true, Linkage::Internal);
    let call_sites = vec![
        indirect_call_through(
            &dispatch,
            4,
            Some(FieldRef {
                record: "ops".into(),
                offset: 8,
            }),
        ),
        indirect_call_through(&call_plain, 5, None),
        indirect_call(
            &bounded,
            6,
            Some(source_location("t.c", 6)),
            Some(vec![h1.id.clone()]),
        ),
        direct_call(&init, &h3, 7),
    ];
    let functions = vec![
        dispatch,
        call_plain,
        bounded,
        init,
        h1,
        h3,
        function("mod_a", "helper", true, Linkage::Internal),
        function("mod_b", "helper", true, Linkage::Internal),
    ];
    let mut base = facts(functions, call_sites);
    base.modules = [("mod_a", "aa"), ("mod_b", "bb")]
        .into_iter()
        .map(|(id, hash)| ModuleReport {
            content_sha256: Some(hash.into()),
            ..report(id, ModuleAnalysis::Analyzed)
        })
        .collect();
    base
}

/// One module the loader verified and extraction never saw, so
/// `analysis.verified` is the only non-zero count.
pub(crate) fn facts_with_one_verified_module() -> ProgramFacts {
    let mut base = facts(vec![], vec![]);
    base.scope.total_entries = 1;
    base.scope.selected_entries = 1;
    base.modules = vec![report("v", ModuleAnalysis::Verified)];
    base
}

/// The header line an overlay file for `session` starts with, for a test
/// that writes the file's records by hand.
pub(crate) fn overlay_header(session: &Session) -> String {
    format!(
        "{{\"v\":{},\"fingerprint\":\"{}\",\"facts\":{}}}\n",
        crate::overlay::OVERLAY_VERSION,
        crate::overlay::fingerprint(session).unwrap(),
        crate::cache::FACTS_FORMAT
    )
}