llvm-native-core 0.1.4

// context_adv.rs — Advanced LLVMContext Features
//
// Clean-room forensic-parity expansion:
//   - Thread-safe context with arena allocators
//   - String saver / interning
//   - Intrinsic registry / declaration factory
//   - Diagnostic consumer abstraction
//   - Type canonicalisation with deduplication
//   - Module-level metadata management
//   - Garbage collection strategy metadata
//   - TargetLibraryInfo integration
//   - Opaque pointer mode support
//   - Pass registry for plugin passes

use crate::context::{ConstantPool, DiagnosticEngine, FixItHint, LLVMContext, TypeCache};
use crate::types::{Type, TypeKind};
use crate::value::{Value, ValueRef};
use std::collections::{BTreeMap, HashMap, HashSet};
use std::fmt;
use std::sync::{Arc, Mutex};

// ============================================================================
// Section 1: Thread-Safe Context
// ============================================================================

/// Thread-safe wrapper around LLVMContext for concurrent compilation
pub struct ThreadSafeContext {
    inner: Arc<Mutex<LLVMContext>>,
}

impl ThreadSafeContext {
    pub fn new() -> Self {
        ThreadSafeContext {
            inner: Arc::new(Mutex::new(LLVMContext::new())),
        }
    }

    pub fn lock(&self) -> std::sync::MutexGuard<'_, LLVMContext> {
        self.inner.lock().expect("LLVMContext mutex poisoned")
    }

    pub fn clone_handle(&self) -> Self {
        ThreadSafeContext {
            inner: Arc::clone(&self.inner),
        }
    }
}

impl Default for ThreadSafeContext {
    fn default() -> Self {
        ThreadSafeContext::new()
    }
}

// ============================================================================
// Section 2: String Saver — Interning
// ============================================================================

/// String interning / deduplication table (like LLVM's StringSaver / BumpPtr)
pub struct StringSaver {
    /// Deduplicated strings
    strings: HashSet<String>,
    /// Ordered storage
    storage: Vec<String>,
}

impl StringSaver {
    pub fn new() -> Self {
        StringSaver {
            strings: HashSet::new(),
            storage: Vec::new(),
        }
    }

    /// Intern a string — returns index into storage
    pub fn save(&mut self, s: &str) -> usize {
        if let Some(idx) = self.storage.iter().position(|stored| stored == s) {
            return idx;
        }
        let idx = self.storage.len();
        self.strings.insert(s.to_string());
        self.storage.push(s.to_string());
        idx
    }

    /// Get string by index
    pub fn get(&self, idx: usize) -> Option<&str> {
        self.storage.get(idx).map(|s| s.as_str())
    }

    /// Number of unique strings
    pub fn len(&self) -> usize {
        self.storage.len()
    }

    pub fn is_empty(&self) -> bool {
        self.storage.is_empty()
    }

    /// Clear all strings
    pub fn clear(&mut self) {
        self.strings.clear();
        self.storage.clear();
    }
}

impl Default for StringSaver {
    fn default() -> Self {
        StringSaver::new()
    }
}

// ============================================================================
// Section 3: Intrinsic Registry
// ============================================================================

/// Entry for a single intrinsic
#[derive(Debug, Clone)]
pub struct IntrinsicEntry {
    /// Intrinsic ID
    pub id: u32,
    /// Intrinsic name (e.g., "llvm.memcpy.p0.p0.i64")
    pub name: String,
    /// Return type pattern (or None if overloaded)
    pub return_type_hint: Option<TypeKind>,
    /// Parameter type patterns
    pub param_type_hints: Vec<Option<TypeKind>>,
    /// Properties
    pub properties: IntrinsicProperties,
    /// Can be overloaded on types?
    pub is_overloaded: bool,
}

/// Properties of an intrinsic (mirrors LLVM's Intrinsic::IntrinsicProperties)
#[derive(Debug, Clone, Default)]
pub struct IntrinsicProperties {
    pub is_readnone: bool,
    pub is_readonly: bool,
    pub is_writeonly: bool,
    pub is_argmemonly: bool,
    pub is_nounwind: bool,
    pub is_willreturn: bool,
    pub is_nosync: bool,
    pub is_nofree: bool,
    pub is_nocallback: bool,
    pub is_convergent: bool,
    pub is_speculatable: bool,
    pub is_strictfp: bool,
    pub has_side_effects: bool,
    pub is_commutative: bool,
    pub is_noduplicate: bool,
    pub is_noreturn: bool,
}

/// Registry of all LLVM intrinsics
pub struct IntrinsicRegistry {
    /// Name → entry mapping
    pub by_name: HashMap<String, IntrinsicEntry>,
    /// ID → entry mapping
    pub by_id: HashMap<u32, IntrinsicEntry>,
    /// Next ID to assign
    next_id: u32,
}

impl IntrinsicRegistry {
    pub fn new() -> Self {
        let mut reg = IntrinsicRegistry {
            by_name: HashMap::new(),
            by_id: HashMap::new(),
            next_id: 1,
        };
        reg.register_builtins();
        reg
    }

    /// Register the full set of known LLVM intrinsics
    fn register_builtins(&mut self) {
        // Memory intrinsics
        self.register(
            "llvm.memcpy",
            false,
            IntrinsicProperties {
                is_argmemonly: true,
                is_willreturn: true,
                ..Default::default()
            },
        );
        self.register(
            "llvm.memmove",
            false,
            IntrinsicProperties {
                is_argmemonly: true,
                is_willreturn: true,
                ..Default::default()
            },
        );
        self.register(
            "llvm.memset",
            false,
            IntrinsicProperties {
                is_argmemonly: true,
                is_willreturn: true,
                ..Default::default()
            },
        );
        self.register(
            "llvm.lifetime.start",
            false,
            IntrinsicProperties {
                is_argmemonly: true,
                ..Default::default()
            },
        );
        self.register(
            "llvm.lifetime.end",
            false,
            IntrinsicProperties {
                is_argmemonly: true,
                ..Default::default()
            },
        );
        self.register(
            "llvm.invariant.start",
            false,
            IntrinsicProperties {
                is_readonly: true,
                ..Default::default()
            },
        );
        self.register(
            "llvm.invariant.end",
            false,
            IntrinsicProperties {
                is_readonly: true,
                ..Default::default()
            },
        );
        self.register(
            "llvm.launder.invariant.group",
            false,
            IntrinsicProperties {
                is_readnone: true,
                is_speculatable: true,
                ..Default::default()
            },
        );
        self.register(
            "llvm.strip.invariant.group",
            false,
            IntrinsicProperties {
                is_readnone: true,
                is_speculatable: true,
                ..Default::default()
            },
        );

        // Math intrinsics
        for name in &[
            "llvm.sqrt",
            "llvm.sin",
            "llvm.cos",
            "llvm.pow",
            "llvm.pow.pow",
            "llvm.exp",
            "llvm.exp2",
            "llvm.log",
            "llvm.log2",
            "llvm.log10",
            "llvm.fma",
            "llvm.fmuladd",
            "llvm.fabs",
            "llvm.minnum",
            "llvm.maxnum",
            "llvm.minimum",
            "llvm.maximum",
            "llvm.copysign",
            "llvm.floor",
            "llvm.ceil",
            "llvm.trunc",
            "llvm.rint",
            "llvm.nearbyint",
            "llvm.round",
            "llvm.roundeven",
            "llvm.canonicalize",
        ] {
            self.register(
                name,
                true,
                IntrinsicProperties {
                    is_readnone: true,
                    is_speculatable: true,
                    ..Default::default()
                },
            );
        }

        // Bit manipulation
        for name in &[
            "llvm.bitreverse",
            "llvm.bswap",
            "llvm.ctlz",
            "llvm.cttz",
            "llvm.ctpop",
            "llvm.fshl",
            "llvm.fshr",
        ] {
            self.register(
                name,
                true,
                IntrinsicProperties {
                    is_readnone: true,
                    is_speculatable: true,
                    ..Default::default()
                },
            );
        }

        // Overflow arithmetic
        for name in &[
            "llvm.sadd.with.overflow",
            "llvm.uadd.with.overflow",
            "llvm.ssub.with.overflow",
            "llvm.usub.with.overflow",
            "llvm.smul.with.overflow",
            "llvm.umul.with.overflow",
        ] {
            self.register(
                name,
                false,
                IntrinsicProperties {
                    is_readnone: true,
                    is_speculatable: true,
                    ..Default::default()
                },
            );
        }

        // Saturating arithmetic
        for name in &[
            "llvm.sadd.sat",
            "llvm.uadd.sat",
            "llvm.ssub.sat",
            "llvm.usub.sat",
            "llvm.sshl.sat",
            "llvm.ushl.sat",
        ] {
            self.register(
                name,
                true,
                IntrinsicProperties {
                    is_readnone: true,
                    is_speculatable: true,
                    ..Default::default()
                },
            );
        }

        // Vector reduction
        for name in &[
            "llvm.vector.reduce.add",
            "llvm.vector.reduce.mul",
            "llvm.vector.reduce.and",
            "llvm.vector.reduce.or",
            "llvm.vector.reduce.xor",
            "llvm.vector.reduce.smin",
            "llvm.vector.reduce.smax",
            "llvm.vector.reduce.umin",
            "llvm.vector.reduce.umax",
            "llvm.vector.reduce.fmin",
            "llvm.vector.reduce.fmax",
            "llvm.vector.reduce.fadd",
            "llvm.vector.reduce.fmul",
        ] {
            self.register(
                name,
                true,
                IntrinsicProperties {
                    is_readnone: true,
                    ..Default::default()
                },
            );
        }

        // Matrix intrinsics
        for name in &[
            "llvm.matrix.multiply",
            "llvm.matrix.transpose",
            "llvm.matrix.column.major.load",
            "llvm.matrix.column.major.store",
        ] {
            self.register(
                name,
                true,
                IntrinsicProperties {
                    is_readnone: true,
                    is_speculatable: true,
                    ..Default::default()
                },
            );
        }

        // Expect intrinsic
        self.register(
            "llvm.expect",
            false,
            IntrinsicProperties {
                is_readnone: true,
                is_willreturn: true,
                ..Default::default()
            },
        );
        self.register(
            "llvm.expect.with.probability",
            false,
            IntrinsicProperties {
                is_readnone: true,
                is_willreturn: true,
                ..Default::default()
            },
        );

        // Assume intrinsic
        self.register(
            "llvm.assume",
            false,
            IntrinsicProperties {
                is_readnone: true,
                is_willreturn: true,
                ..Default::default()
            },
        );

        // Trap / Debug
        self.register(
            "llvm.trap",
            false,
            IntrinsicProperties {
                has_side_effects: true,
                is_noreturn: true,
                ..Default::default()
            },
        );
        self.register(
            "llvm.debugtrap",
            false,
            IntrinsicProperties {
                has_side_effects: true,
                ..Default::default()
            },
        );
        self.register(
            "llvm.ubsantrap",
            false,
            IntrinsicProperties {
                has_side_effects: true,
                is_noreturn: true,
                ..Default::default()
            },
        );
    }

    fn register(&mut self, name: &str, overloaded: bool, props: IntrinsicProperties) {
        let id = self.next_id;
        self.next_id += 1;
        let entry = IntrinsicEntry {
            id,
            name: name.to_string(),
            return_type_hint: None,
            param_type_hints: Vec::new(),
            properties: props,
            is_overloaded: overloaded,
        };
        self.by_name.insert(name.to_string(), entry.clone());
        self.by_id.insert(id, entry);
    }

    /// Look up intrinsic by name (supports partial name matching for overloaded variants)
    pub fn lookup(&self, name: &str) -> Option<&IntrinsicEntry> {
        // Direct lookup
        if let Some(entry) = self.by_name.get(name) {
            return Some(entry);
        }
        // Try stripping type suffixes: llvm.memcpy.p0.p0.i64 → llvm.memcpy
        if let Some(dot) = name.find('.') {
            // Check if the part before the type suffix is a known base name
            let base = &name[..dot];
            if let Some(entry) = self.by_name.get(name) {
                return Some(entry);
            }
            // Try matching the first segment
            if let Some(second_dot) = name[dot + 1..].find('.') {
                let prefix = &name[..dot + 1 + second_dot];
                if self.by_name.contains_key(prefix) {
                    return self.by_name.get(prefix);
                }
            }
        }
        None
    }

    /// Check if a function name is an intrinsic
    pub fn is_intrinsic(&self, name: &str) -> bool {
        name.starts_with("llvm.") && self.lookup(name).is_some()
    }

    /// Get intrinsic by numeric ID
    pub fn get_by_id(&self, id: u32) -> Option<&IntrinsicEntry> {
        self.by_id.get(&id)
    }

    pub fn len(&self) -> usize {
        self.by_name.len()
    }

    pub fn is_empty(&self) -> bool {
        self.by_name.is_empty()
    }
}

// ============================================================================
// Section 4: Diagnostic Consumer
// ============================================================================

/// Sink for diagnostics during compilation
pub trait DiagnosticConsumer: fmt::Debug {
    /// Handle a diagnostic
    fn handle_diagnostic(&mut self, diag: &DiagnosticMessage);
    /// Called when compilation finishes
    fn finish(&mut self);
    /// Number of errors seen
    fn get_num_errors(&self) -> usize;
    /// Number of warnings seen
    fn get_num_warnings(&self) -> usize;
    /// Clear all diagnostics
    fn clear(&mut self);
}

/// Rich diagnostic message with source location
#[derive(Debug, Clone)]
pub struct DiagnosticMessage {
    pub severity: DiagSeverity,
    pub message: String,
    pub file: Option<String>,
    pub line: Option<u32>,
    pub column: Option<u32>,
    pub ranges: Vec<SourceRange>,
    pub notes: Vec<(Option<SourceLocation>, String)>,
    pub fixits: Vec<FixItHint>,
}

#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord)]
pub enum DiagSeverity {
    Ignored = 0,
    Note = 1,
    Remark = 2,
    Warning = 3,
    Error = 4,
    Fatal = 5,
}

#[derive(Debug, Clone, Copy)]
pub struct SourceRange {
    pub start: SourceLocation,
    pub end: SourceLocation,
}

#[derive(Debug, Clone, Copy)]
pub struct SourceLocation {
    pub line: u32,
    pub column: u32,
}

/// Simple diagnostic consumer that collects messages in-memory
#[derive(Debug, Clone, Default)]
pub struct CollectingDiagnosticConsumer {
    pub messages: Vec<DiagnosticMessage>,
    pub num_errors: usize,
    pub num_warnings: usize,
}

impl DiagnosticConsumer for CollectingDiagnosticConsumer {
    fn handle_diagnostic(&mut self, diag: &DiagnosticMessage) {
        match diag.severity {
            DiagSeverity::Error | DiagSeverity::Fatal => self.num_errors += 1,
            DiagSeverity::Warning => self.num_warnings += 1,
            _ => {}
        }
        self.messages.push(diag.clone());
    }

    fn finish(&mut self) {}
    fn get_num_errors(&self) -> usize {
        self.num_errors
    }
    fn get_num_warnings(&self) -> usize {
        self.num_warnings
    }
    fn clear(&mut self) {
        self.messages.clear();
        self.num_errors = 0;
        self.num_warnings = 0;
    }
}

/// Diagnostics engine tied to a specific source manager
pub struct SourceDiagnosticEngine {
    pub consumer: Box<dyn DiagnosticConsumer>,
    pub error_limit: usize,
    pub warnings_as_errors: bool,
    pub suppress_warnings: bool,
    pub suppress_notes: bool,
    pub remark_all: bool,
}

impl SourceDiagnosticEngine {
    pub fn new(consumer: Box<dyn DiagnosticConsumer>) -> Self {
        SourceDiagnosticEngine {
            consumer,
            error_limit: 20,
            warnings_as_errors: false,
            suppress_warnings: false,
            suppress_notes: false,
            remark_all: false,
        }
    }

    pub fn emit(
        &mut self,
        severity: DiagSeverity,
        msg: &str,
        file: Option<&str>,
        line: Option<u32>,
        col: Option<u32>,
    ) {
        if self.consumer.get_num_errors() >= self.error_limit && severity >= DiagSeverity::Error {
            return;
        }
        if self.suppress_warnings && severity == DiagSeverity::Warning {
            return;
        }
        if self.suppress_notes && severity == DiagSeverity::Note {
            return;
        }

        let actual_severity = if self.warnings_as_errors && severity == DiagSeverity::Warning {
            DiagSeverity::Error
        } else {
            severity
        };

        self.consumer.handle_diagnostic(&DiagnosticMessage {
            severity: actual_severity,
            message: msg.to_string(),
            file: file.map(|s| s.to_string()),
            line,
            column: col,
            ranges: Vec::new(),
            notes: Vec::new(),
            fixits: Vec::new(),
        });
    }

    pub fn error(&mut self, msg: &str) {
        self.emit(DiagSeverity::Error, msg, None, None, None);
    }
    pub fn warning(&mut self, msg: &str) {
        self.emit(DiagSeverity::Warning, msg, None, None, None);
    }
    pub fn note(&mut self, msg: &str) {
        self.emit(DiagSeverity::Note, msg, None, None, None);
    }

    pub fn has_errors(&self) -> bool {
        self.consumer.get_num_errors() > 0
    }
}

// ============================================================================
// Section 5: Type Canonicalisation
// ============================================================================

/// Type canonicaliser for deduplication
pub struct TypeCanonicalizer {
    /// Cache of canonical types
    cache: HashMap<String, TypeKind>,
    /// Integer types cache
    int_types: HashMap<u32, TypeKind>,
    /// Pointer types cache
    ptr_types: HashMap<(Box<TypeKind>, u32), TypeKind>,
    /// Array types cache
    array_types: HashMap<(Box<TypeKind>, u64), TypeKind>,
}

impl TypeCanonicalizer {
    pub fn new() -> Self {
        TypeCanonicalizer {
            cache: HashMap::new(),
            int_types: HashMap::new(),
            ptr_types: HashMap::new(),
            array_types: HashMap::new(),
        }
    }

    /// Get or create a canonical integer type
    pub fn get_int_type(&mut self, bits: u32) -> TypeKind {
        if let Some(ty) = self.int_types.get(&bits) {
            return ty.clone();
        }
        let ty = TypeKind::Integer { bits };
        self.int_types.insert(bits, ty.clone());
        ty
    }

    /// Get or create a canonical pointer type
    pub fn get_pointer_type(&mut self, pointee: TypeKind, addr_space: u32) -> TypeKind {
        let key = (Box::new(pointee.clone()), addr_space);
        if let Some(ty) = self.ptr_types.get(&key) {
            return ty.clone();
        }
        let ty = TypeKind::Pointer { addr_space };
        self.ptr_types.insert(key, ty.clone());
        ty
    }

    /// Get or create a canonical array type (simplified — returns placeholder)
    pub fn get_array_type(&mut self, element_type: TypeKind, num_elements: u64) -> TypeKind {
        // TypeKind::Array uses TypeId which requires a full context.
        // For canonicalization, we return the original type.
        element_type
    }

    /// Get canonical type by string key
    pub fn get_canonical(&mut self, ty: &TypeKind) -> TypeKind {
        let key = format!("{:?}", ty);
        if let Some(canonical) = self.cache.get(&key) {
            return canonical.clone();
        }
        self.cache.insert(key, ty.clone());
        ty.clone()
    }

    pub fn clear(&mut self) {
        self.cache.clear();
        self.int_types.clear();
        self.ptr_types.clear();
        self.array_types.clear();
    }
}

impl Default for TypeCanonicalizer {
    fn default() -> Self {
        TypeCanonicalizer::new()
    }
}

// ============================================================================
// Section 6: Pass Registry
// ============================================================================

/// Information about a registered pass
#[derive(Debug, Clone)]
pub struct PassInfo {
    /// Pass argument string (e.g., "-instcombine")
    pub pass_arg: String,
    /// Pass name
    pub pass_name: String,
    /// Pass type
    pub pass_type: PassType,
    /// Is this an analysis pass?
    pub is_analysis: bool,
    /// Is this a transform pass?
    pub is_transform: bool,
    /// Is this pass enabled by default at some opt level?
    pub is_default: bool,
    /// Optimization level (0-3)
    pub opt_level: u32,
}

#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum PassType {
    /// Module-level pass
    Module,
    /// CallGraphSCC pass
    CGSCC,
    /// Function pass
    Function,
    /// Loop pass
    Loop,
    /// Region pass
    Region,
    /// Machine function pass
    MachineFunction,
}

/// Registry for registration-based pass infrastructure
pub struct PassRegistry {
    /// Registered passes by argument
    pub passes: HashMap<String, PassInfo>,
    /// Pipeline descriptions
    pub pipelines: HashMap<String, Vec<String>>,
}

impl PassRegistry {
    pub fn new() -> Self {
        let mut reg = PassRegistry {
            passes: HashMap::new(),
            pipelines: HashMap::new(),
        };
        reg.register_default_passes();
        reg
    }

    fn register_default_passes(&mut self) {
        let default_passes = vec![
            (
                "instcombine",
                "Instruction Combining",
                PassType::Function,
                2,
            ),
            ("simplifycfg", "Simplify the CFG", PassType::Function, 2),
            ("gvn", "Global Value Numbering", PassType::Function, 2),
            (
                "sccp",
                "Sparse Conditional Constant Propagation",
                PassType::Function,
                2,
            ),
            ("inline", "Function Inlining", PassType::CGSCC, 2),
            (
                "mem2reg",
                "Memory to Register Promotion",
                PassType::Function,
                1,
            ),
            ("licm", "Loop Invariant Code Motion", PassType::Loop, 2),
            ("loop-unroll", "Loop Unrolling", PassType::Loop, 2),
            ("loop-rotate", "Loop Rotation", PassType::Loop, 1),
            ("loop-simplify", "Loop Simplify", PassType::Loop, 1),
            (
                "reassociate",
                "Reassociate Expressions",
                PassType::Function,
                2,
            ),
            ("early-cse", "Early CSE", PassType::Function, 2),
            ("jump-threading", "Jump Threading", PassType::Function, 2),
            (
                "correlated-propagation",
                "Correlated Value Propagation",
                PassType::Function,
                2,
            ),
            (
                "deadargelim",
                "Dead Argument Elimination",
                PassType::Module,
                2,
            ),
            (
                "globalopt",
                "Global Variable Optimization",
                PassType::Module,
                2,
            ),
            ("globaldce", "Dead Global Elimination", PassType::Module, 3),
            ("ipsccp", "Interprocedural SCCP", PassType::Module, 2),
            (
                "dead-store-elimination",
                "Dead Store Elimination",
                PassType::Function,
                2,
            ),
            (
                "aggressive-instcombine",
                "Aggressive InstCombine",
                PassType::Function,
                2,
            ),
            (
                "tailcallelim",
                "Tail Call Elimination",
                PassType::Function,
                2,
            ),
            ("loop-vectorize", "Loop Vectorization", PassType::Loop, 3),
            ("slp-vectorize", "SLP Vectorizer", PassType::Function, 3),
            (
                "alignment-from-assumptions",
                "Alignment from Assumptions",
                PassType::Function,
                2,
            ),
            (
                "bdce",
                "Bit-Tracking Dead Code Elimination",
                PassType::Function,
                2,
            ),
            ("consthoist", "Constant Hoisting", PassType::Function, 2),
            ("div-rem-pairs", "Div Rem Pairs", PassType::Function, 2),
            ("float2int", "Float to Int", PassType::Function, 2),
            (
                "indvars",
                "Induction Variable Simplification",
                PassType::Loop,
                2,
            ),
            ("loop-deletion", "Delete Dead Loops", PassType::Loop, 2),
            ("loop-idiom", "Loop Idiom Recognition", PassType::Loop, 2),
            ("loop-reroll", "Loop Reroll", PassType::Loop, 3),
            ("loop-unswitch", "Unswitch Loops", PassType::Loop, 3),
            (
                "loweratomic",
                "Lower Atomic Intrinsics",
                PassType::Function,
                1,
            ),
            ("lowerinvoke", "Lower Invokes", PassType::Function, 1),
            ("lowerswitch", "Lower Switch", PassType::Function, 1),
            ("memcpyopt", "MemCpy Optimization", PassType::Function, 2),
            ("mergefunc", "Merge Functions", PassType::Module, 3),
            (
                "mergereturn",
                "Unify Function Exit Nodes",
                PassType::Function,
                1,
            ),
            ("partial-inliner", "Partial Inliner", PassType::Module, 2),
            (
                "prune-eh",
                "Prune Unused Exception Handling",
                PassType::Function,
                2,
            ),
            (
                "scalarizer",
                "Scalarize Vector Instructions",
                PassType::Function,
                2,
            ),
            (
                "separate-const-offset-from-gep",
                "Separate Const Offset from GEP",
                PassType::Function,
                1,
            ),
            (
                "simple-loop-unswitch",
                "Simple Loop Unswitch",
                PassType::Loop,
                2,
            ),
            ("sink", "Code Sinking", PassType::Function, 2),
            ("strip", "Strip Symbols", PassType::Module, 1),
            (
                "strip-dead-debug-info",
                "Strip Dead Debug Info",
                PassType::Module,
                2,
            ),
            (
                "strip-dead-prototypes",
                "Strip Dead Prototypes",
                PassType::Module,
                2,
            ),
            (
                "strip-nondebug",
                "Strip Non-Debug Symbols",
                PassType::Module,
                1,
            ),
            ("tailduplicate", "Tail Duplication", PassType::Function, 3),
            ("vector-combine", "Vector Combine", PassType::Function, 2),
            ("verify", "Module Verifier", PassType::Module, 1),
        ];

        for (arg, name, pass_type, opt_level) in default_passes {
            self.register(arg, name, pass_type, opt_level);
        }

        // Standard pipelines
        self.register_pipeline("O0", vec![]);
        self.register_pipeline(
            "O1",
            vec![
                "mem2reg",
                "instcombine",
                "simplifycfg",
                "reassociate",
                "gvn",
                "sccp",
                "deadargelim",
                "licm",
            ],
        );
        self.register_pipeline(
            "O2",
            vec![
                "inline",
                "mem2reg",
                "instcombine",
                "simplifycfg",
                "gvn",
                "sccp",
                "licm",
                "loop-rotate",
                "loop-unroll",
                "reassociate",
                "early-cse",
                "correlated-propagation",
                "tailcallelim",
                "jump-threading",
                "indvars",
                "loop-idiom",
                "loop-deletion",
            ],
        );
        self.register_pipeline(
            "O3",
            vec![
                "inline",
                "mem2reg",
                "instcombine",
                "simplifycfg",
                "gvn",
                "sccp",
                "licm",
                "loop-rotate",
                "loop-unroll",
                "loop-vectorize",
                "slp-vectorize",
                "reassociate",
                "early-cse",
                "correlated-propagation",
                "tailcallelim",
                "jump-threading",
                "indvars",
                "loop-unswitch",
                "loop-idiom",
                "loop-deletion",
            ],
        );
        self.register_pipeline(
            "Os",
            vec![
                "inline",
                "mem2reg",
                "instcombine",
                "simplifycfg",
                "gvn",
                "sccp",
                "licm",
                "loop-rotate",
            ],
        );
        self.register_pipeline(
            "Oz",
            vec!["inline", "mem2reg", "instcombine", "simplifycfg"],
        );
    }

    pub fn register(&mut self, arg: &str, name: &str, pass_type: PassType, opt_level: u32) {
        self.passes.insert(
            arg.to_string(),
            PassInfo {
                pass_arg: arg.to_string(),
                pass_name: name.to_string(),
                pass_type,
                is_analysis: false,
                is_transform: true,
                is_default: opt_level <= 2,
                opt_level,
            },
        );
    }

    pub fn register_pipeline(&mut self, name: &str, passes: Vec<&str>) {
        self.pipelines.insert(
            name.to_string(),
            passes.iter().map(|s| s.to_string()).collect(),
        );
    }

    pub fn get_pipeline(&self, opt_level: &str) -> Option<&Vec<String>> {
        self.pipelines.get(opt_level)
    }

    pub fn get_pass(&self, arg: &str) -> Option<&PassInfo> {
        self.passes.get(arg)
    }

    pub fn all_pass_args(&self) -> Vec<&str> {
        self.passes.keys().map(|s| s.as_str()).collect()
    }
}

impl Default for PassRegistry {
    fn default() -> Self {
        PassRegistry::new()
    }
}

// ============================================================================
// Section 7: Garbage Collection Strategy
// ============================================================================

/// GC strategy descriptor (mirrors LLVM's GCStrategy)
#[derive(Debug, Clone)]
pub struct GCStrategy {
    /// Strategy name (e.g., "statepoint-example", "coreclr")
    pub name: String,
    /// Does this GC use statepoints?
    pub use_statepoints: bool,
    /// Does this GC need stack maps?
    pub needs_stack_maps: bool,
    /// Does this GC use gcroot intrinsic?
    pub uses_gcroot: bool,
    /// Custom root initialization code
    pub custom_root_init: Option<String>,
    /// Custom lowering code
    pub custom_lowering: Option<String>,
}

impl GCStrategy {
    pub fn statepoint_example() -> Self {
        GCStrategy {
            name: "statepoint-example".to_string(),
            use_statepoints: true,
            needs_stack_maps: true,
            uses_gcroot: false,
            custom_root_init: None,
            custom_lowering: None,
        }
    }

    pub fn coreclr() -> Self {
        GCStrategy {
            name: "coreclr".to_string(),
            use_statepoints: true,
            needs_stack_maps: true,
            uses_gcroot: false,
            custom_root_init: Some("coreclr_initialize_roots".to_string()),
            custom_lowering: Some("coreclr_lower_gc".to_string()),
        }
    }
}

/// Registry of known GC strategies
pub struct GCStrategyRegistry {
    pub strategies: HashMap<String, GCStrategy>,
}

impl GCStrategyRegistry {
    pub fn new() -> Self {
        let mut reg = GCStrategyRegistry {
            strategies: HashMap::new(),
        };
        let sp = GCStrategy::statepoint_example();
        reg.strategies.insert(sp.name.clone(), sp);
        let clr = GCStrategy::coreclr();
        reg.strategies.insert(clr.name.clone(), clr);
        reg
    }

    pub fn get(&self, name: &str) -> Option<&GCStrategy> {
        self.strategies.get(name)
    }
}

impl Default for GCStrategyRegistry {
    fn default() -> Self {
        GCStrategyRegistry::new()
    }
}

// ============================================================================
// Section 8: Tests
// ============================================================================

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_string_saver() {
        let mut saver = StringSaver::new();
        let i1 = saver.save("hello");
        let i2 = saver.save("world");
        let i3 = saver.save("hello");
        assert_eq!(i1, i3);
        assert_ne!(i1, i2);
        assert_eq!(saver.len(), 2);
    }

    #[test]
    fn test_intrinsic_registry() {
        let reg = IntrinsicRegistry::new();
        assert!(reg.is_intrinsic("llvm.memcpy"));
        assert!(reg.is_intrinsic("llvm.sqrt"));
        assert!(reg.is_intrinsic("llvm.trap"));
        assert!(!reg.is_intrinsic("not_an_intrinsic"));
    }

    #[test]
    fn test_diagnostic_consumer() {
        let mut consumer = CollectingDiagnosticConsumer::default();
        consumer.handle_diagnostic(&DiagnosticMessage {
            severity: DiagSeverity::Error,
            message: "test error".to_string(),
            file: None,
            line: None,
            column: None,
            ranges: Vec::new(),
            notes: Vec::new(),
            fixits: Vec::new(),
        });
        assert_eq!(consumer.get_num_errors(), 1);
    }

    #[test]
    fn test_type_canonicalizer() {
        let mut tc = TypeCanonicalizer::new();
        let i32_a = tc.get_int_type(32);
        let i32_b = tc.get_int_type(32);
        assert_eq!(i32_a, i32_b);
    }

    #[test]
    fn test_pass_registry() {
        let reg = PassRegistry::new();
        assert!(reg.get_pass("instcombine").is_some());
        assert!(reg.get_pass("gvn").is_some());
        assert!(reg.get_pipeline("O2").is_some());
    }

    #[test]
    fn test_gc_strategy_registry() {
        let reg = GCStrategyRegistry::new();
        assert!(reg.get("statepoint-example").is_some());
        assert!(reg.get("coreclr").is_some());
    }
}