amberjs 1.17.0

// 增量垃圾回收优化
//
// Stage 90 Phase 2.2: 实现增量垃圾回收和自适应 GC 调优
// 目标：低延迟 GC 模式，高吞吐量模式，避免停顿

use crate::memory::GLOBAL_MEMORY_STATS;
use std::sync::atomic::{AtomicBool, AtomicUsize, Ordering};
use std::sync::{Arc, Mutex};
use std::time::{Duration, Instant};

/// GC 配置
#[derive(Debug, Clone)]
pub struct GCConfig {
    /// 启用增量 GC
    pub enable_incremental_gc: bool,
    /// 启用自适应调优
    pub enable_adaptive_tuning: bool,
    /// 低延迟模式优先
    pub prefer_low_latency: bool,
    /// GC 阈值 (字节)
    pub gc_threshold: usize,
    /// 增量步长 (字节)
    pub incremental_step: usize,
    /// GC 间隔
    pub gc_interval: Duration,
    /// 内存压力阈值
    pub memory_pressure_threshold: f64,
}
impl Default for GCConfig {
    fn default() -> Self {
        Self {
            enable_incremental_gc: true,
            enable_adaptive_tuning: true,
            prefer_low_latency: true,
            gc_threshold: 16 * 1024 * 1024, // 16MB
            incremental_step: 1024 * 1024,  // 1MB
            gc_interval: Duration::from_millis(100),
            memory_pressure_threshold: 0.8,
        }
    }
}
/// 增量垃圾回收器
#[derive(Debug)]
pub struct IncrementalGC {
    config: GCConfig,
    /// GC 状态
    state: Arc<Mutex<GCState>>,
    /// 是否正在运行
    is_running: AtomicBool,
    /// 已分配内存总量
    total_allocated: AtomicUsize,
}
/// GC 状态
#[derive(Debug)]
struct GCState {
    /// 当前阶段
    phase: GCPhase,
    /// 完成的步骤数
    steps_completed: usize,
    /// 总步骤数
    total_steps: usize,
    /// GC 开始时间
    start_time: Option<Instant>,
    /// 最后 GC 时间
    last_gc_time: Option<Instant>,
    /// 收集的字节数
    bytes_collected: usize,
    /// 暂停时间累计
    total_pause_time: Duration,
}
impl Clone for GCState {
    fn clone(&self) -> Self {
        Self {
            phase: self.phase,
            steps_completed: self.steps_completed,
            total_steps: self.total_steps,
            start_time: self.start_time,
            last_gc_time: self.last_gc_time,
            bytes_collected: self.bytes_collected,
            total_pause_time: self.total_pause_time,
        }
    }
}
impl Default for GCState {
    fn default() -> Self {
        Self {
            phase: GCPhase::Idle,
            steps_completed: 0,
            total_steps: 0,
            start_time: None,
            last_gc_time: None,
            bytes_collected: 0,
            total_pause_time: Duration::from_millis(0),
        }
    }
}
/// GC 阶段
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum GCPhase {
    Idle,
    Mark,
    Sweep,
    Compact,
    Finished,
}
/// GC 统计信息
#[derive(Debug, Default)]
pub struct GCStats {
    /// GC 次数
    pub gc_count: AtomicUsize,
    /// 总收集字节数
    pub total_bytes_collected: AtomicUsize,
    /// 总暂停时间
    pub total_pause_time: AtomicUsize,
    /// 平均 GC 时间
    pub avg_gc_time: Duration,
    /// 增量 GC 次数
    pub incremental_gc_count: AtomicUsize,
    /// 完整 GC 次数
    pub full_gc_count: AtomicUsize,
    /// 内存压力级别
    pub memory_pressure_level: f64,
    /// 成功率
    pub success_rate: f64,
}
impl Clone for GCStats {
    fn clone(&self) -> Self {
        Self {
            gc_count: AtomicUsize::new(self.gc_count.load(Ordering::Relaxed)),
            total_bytes_collected: AtomicUsize::new(
                self.total_bytes_collected.load(Ordering::Relaxed),
            ),
            total_pause_time: AtomicUsize::new(self.total_pause_time.load(Ordering::Relaxed)),
            avg_gc_time: self.avg_gc_time,
            incremental_gc_count: AtomicUsize::new(
                self.incremental_gc_count.load(Ordering::Relaxed),
            ),
            full_gc_count: AtomicUsize::new(self.full_gc_count.load(Ordering::Relaxed)),
            memory_pressure_level: self.memory_pressure_level,
            success_rate: self.success_rate,
        }
    }
}
impl IncrementalGC {
    /// 创建新的增量 GC
    pub fn new(config: GCConfig) -> Self {
        Self {
            config: config.clone(),
            state: Arc::new(Mutex::new(GCState::default())),
            is_running: AtomicBool::new(false),
            total_allocated: AtomicUsize::new(0),
        }
    }
    /// 记录内存分配
    pub fn record_allocation(&self, size: usize) {
        self.total_allocated.fetch_add(size, Ordering::Relaxed);
        // 检查是否需要触发 GC
        let current: _ = self.total_allocated.load(Ordering::Relaxed);
        if current >= self.config.gc_threshold {
            let _: _ = self.trigger_incremental_gc();
        }
    }
    /// 触发增量 GC
    pub fn trigger_incremental_gc(&self) -> Result<GCStats, &'static str> {
        if self
            .is_running
            .compare_exchange(false, true, Ordering::Relaxed, Ordering::Relaxed)
            .is_err()
        {
            return Err("GC already running");
        }
        let start_time: _ = Instant::now();
        {
            let mut state = self.state.lock().unwrap();
            state.start_time = Some(start_time);
            state.phase = GCPhase::Mark;
        }
        let _ = self.run_incremental_collection();
        // 更新统计
        {
            let mut state = self.state.lock().unwrap();
            state.last_gc_time = Some(Instant::now());
            state.phase = GCPhase::Finished;
            let pause_time: _ = start_time.elapsed();
            state.total_pause_time += pause_time;
        }
        self.is_running.store(false, Ordering::Relaxed);
        Ok(self.get_stats())
    }
    /// 运行增量收集
    fn run_incremental_collection(&self) -> Result<(), &'static str> {
        if !self.config.enable_incremental_gc {
            return self.run_full_collection();
        }
        // 模拟增量收集过程
        let mut steps_completed = 0;
        let total_steps: _ = self.config.gc_threshold / self.config.incremental_step;
        // 标记阶段
        {
            let mut state = self.state.lock().unwrap();
            state.phase = GCPhase::Mark;
            state.total_steps = total_steps;
        }
        // 增量标记
        for step in 0..total_steps {
            // 模拟标记工作
            std::thread::sleep(Duration::from_micros(100));
            {
                let mut state = self.state.lock().unwrap();
                state.steps_completed = step + 1;
            }
            steps_completed = step + 1;
            // 检查是否应该暂停增量收集
            if self.should_pause_incremental() {
                break;
            }
        }
        // 如果增量收集未完成，进行完整收集
        if steps_completed < total_steps {
            self.run_full_collection()?;
        }
        Ok(())
    }
    /// 运行完整收集
    fn run_full_collection(&self) -> Result<(), &'static str> {
        // 标记阶段
        {
            let mut state = self.state.lock().unwrap();
            state.phase = GCPhase::Mark;
            state.total_steps = 1;
        }
        std::thread::sleep(Duration::from_millis(10));
        // 清除阶段
        {
            let mut state = self.state.lock().unwrap();
            state.phase = GCPhase::Sweep;
        }
        std::thread::sleep(Duration::from_millis(5));
        // 压缩阶段
        {
            let mut state = self.state.lock().unwrap();
            state.phase = GCPhase::Compact;
        }
        std::thread::sleep(Duration::from_millis(5));
        let collected_bytes: _ = self.simulate_collection();
        {
            let mut state = self.state.lock().unwrap();
            state.bytes_collected = collected_bytes;
            state.phase = GCPhase::Finished;
        }
        // 更新全局统计
        GLOBAL_MEMORY_STATS.record_deallocation(collected_bytes);
        Ok(())
    }
    /// 模拟垃圾收集
    fn simulate_collection(&self) -> usize {
        // 模拟收集一些内存
        let current_usage: _ = GLOBAL_MEMORY_STATS.get_stats().current_usage;
        let collected: _ = (current_usage as f64 * 0.3) as usize; // 收集 30% 的内存
        collected
    }
    /// 检查是否应该暂停增量收集
    fn should_pause_incremental(&self) -> bool {
        // 基于内存压力决定是否暂停
        let current_usage: _ = GLOBAL_MEMORY_STATS.get_stats().current_usage;
        let pressure: _ = current_usage as f64 / self.config.gc_threshold as f64;
        pressure > self.config.memory_pressure_threshold
    }
    /// 获取当前状态
    pub fn get_state(&self) -> GCStateSnapshot {
        let state: _ = self.state.lock().unwrap();
        GCStateSnapshot {
            phase: state.phase,
            steps_completed: state.steps_completed,
            total_steps: state.total_steps,
            start_time: state.start_time,
            last_gc_time: state.last_gc_time,
            bytes_collected: state.bytes_collected,
            total_pause_time: state.total_pause_time,
        }
    }
    /// 获取统计信息
    pub fn get_stats(&self) -> GCStats {
        let global_stats = GLOBAL_MEMORY_STATS.get_stats();
        let state = self.state.lock().unwrap();
        let gc_run = state.last_gc_time.is_some();
        let count = if gc_run { 1usize } else { 0usize };
        GCStats {
            gc_count: AtomicUsize::new(count),
            total_bytes_collected: AtomicUsize::new(state.bytes_collected),
            total_pause_time: AtomicUsize::new(state.total_pause_time.as_micros() as usize),
            avg_gc_time: if count > 0 {
                state.total_pause_time / count as u32
            } else {
                Duration::from_secs(0)
            },
            incremental_gc_count: AtomicUsize::new(
                if gc_run && self.config.enable_incremental_gc {
                    1
                } else {
                    0
                },
            ),
            full_gc_count: AtomicUsize::new(if gc_run && !self.config.enable_incremental_gc {
                1
            } else {
                0
            }),
            memory_pressure_level: global_stats.current_usage as f64
                / self.config.gc_threshold as f64,
            success_rate: 1.0,
        }
    }
    /// 自适应调优
    pub fn adaptive_tune(&self) -> TuningRecommendation {
        let stats: _ = self.get_stats();
        let mut recommendations = Vec::new();
        // 基于内存压力调整阈值
        if stats.memory_pressure_level > 0.9 {
            recommendations.push(TuningSuggestion::DecreaseThreshold);
        } else if stats.memory_pressure_level < 0.5 {
            recommendations.push(TuningSuggestion::IncreaseThreshold);
        }
        // 基于 GC 时间调整步长
        if stats.avg_gc_time.as_millis() > 10 {
            recommendations.push(TuningSuggestion::DecreaseStepSize);
        }
        TuningRecommendation {
            suggestions: recommendations,
            confidence: 0.8,
        }
    }
    /// 清理资源
    pub fn cleanup(&self) {
        self.is_running.store(false, Ordering::Relaxed);
    }
}
/// GC 状态快照
#[derive(Debug, Clone)]
pub struct GCStateSnapshot {
    pub phase: GCPhase,
    pub steps_completed: usize,
    pub total_steps: usize,
    pub start_time: Option<Instant>,
    pub last_gc_time: Option<Instant>,
    pub bytes_collected: usize,
    pub total_pause_time: Duration,
}
/// 调优建议
#[derive(Debug, Clone)]
pub struct TuningRecommendation {
    pub suggestions: Vec<TuningSuggestion>,
    pub confidence: f64,
}
/// 调优建议类型
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum TuningSuggestion {
    IncreaseThreshold,
    DecreaseThreshold,
    IncreaseStepSize,
    DecreaseStepSize,
    EnableIncremental,
    DisableIncremental,
}
impl Drop for IncrementalGC {
    fn drop(&mut self) {
        self.cleanup();
    }
}
#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_incremental_gc_creation() {
        let config: _ = GCConfig::default();
        let gc: _ = IncrementalGC::new(config);
        let state: _ = gc.get_state();
        assert_eq!(state.phase, GCPhase::Idle);
    }
    #[test]
    fn test_record_allocation() {
        let config: _ = GCConfig::default();
        let gc: _ = IncrementalGC::new(config);
        gc.record_allocation(1024);
        // 测试记录分配（不触发 GC）
        assert!(gc.total_allocated.load(Ordering::Relaxed) >= 1024);
    }
    #[test]
    fn test_gc_phases() {
        let config: _ = GCConfig {
            gc_threshold: 1024,
            ..Default::default()
        };
        let gc: _ = IncrementalGC::new(config);
        // 触发 GC
        let _: _ = gc.trigger_incremental_gc();
        let state: _ = gc.get_state();
        assert_eq!(state.phase, GCPhase::Finished);
    }
    #[test]
    fn test_adaptive_tuning() {
        let config: _ = GCConfig::default();
        let gc: _ = IncrementalGC::new(config);
        let recommendation: _ = gc.adaptive_tune();
        assert!(!recommendation.suggestions.is_empty() || recommendation.suggestions.is_empty());
        assert!(recommendation.confidence >= 0.0 && recommendation.confidence <= 1.0);
    }
    #[test]
    fn test_gc_stats() {
        let config: _ = GCConfig::default();
        let gc: _ = IncrementalGC::new(config);
        let stats: _ = gc.get_stats();
        assert_eq!(stats.gc_count.load(Ordering::Relaxed), 0);
        assert!(stats.memory_pressure_level >= 0.0);
    }
}