fix(proxy): serialize per-app provider switches to prevent state corruption

Concurrent failover switches for the same app could cause is_current,
local settings, and Live backup to point at different providers.

- Add SwitchLockManager with per-app mutexes (different apps still parallel)
- Unify scattered switch logic into ProxyService::hot_switch_provider
- Fix TOCTOU in set_current_provider via mutate_settings
- Add logical_target_changed to skip redundant UI refreshes
- Add tests for serialization and restore-waits-for-switch scenarios
This commit is contained in:
Jason
2026-03-25 16:02:35 +08:00
parent 6a083cdd1c
commit af8f907467
6 changed files with 344 additions and 95 deletions
+11 -23
View File
@@ -2,15 +2,12 @@
//!
//! 处理故障转移成功后的供应商切换逻辑,包括:
//! - 去重控制(避免多个请求同时触发)
//! - 数据库更新
//! - 托盘菜单更新
//! - 前端事件发射
//! - Live 备份更新
use crate::database::Database;
use crate::error::AppError;
use std::collections::HashSet;
use std::str::FromStr;
use std::sync::Arc;
use tauri::{Emitter, Manager};
use tokio::sync::RwLock;
@@ -98,30 +95,21 @@ impl FailoverSwitchManager {
log::info!("[FO-001] 切换: {app_type} → {provider_name}");
// 1. 更新数据库 is_current
self.db.set_current_provider(app_type, provider_id)?;
let mut switched = false;
// 2. 更新本地 settings(设备级)
let app_type_enum = crate::app_config::AppType::from_str(app_type)
.map_err(|_| AppError::Message(format!("无效的应用类型: {app_type}")))?;
crate::settings::set_current_provider(&app_type_enum, Some(provider_id))?;
// 3. 更新托盘菜单和发射事件
if let Some(app) = app_handle {
// 更新托盘菜单
if let Some(app_state) = app.try_state::<crate::store::AppState>() {
// 更新 Live 备份(确保代理停止时恢复正确配置)
if let Ok(Some(provider)) = self.db.get_provider_by_id(provider_id, app_type) {
if let Err(e) = app_state
.proxy_service
.update_live_backup_from_provider(app_type, &provider)
.await
{
log::warn!("[FO-003] Live 备份更新失败: {e}");
}
switched = app_state
.proxy_service
.hot_switch_provider(app_type, provider_id)
.await
.map_err(AppError::Message)?
.logical_target_changed;
if !switched {
return Ok(false);
}
// 重建托盘菜单
if let Ok(new_menu) = crate::tray::create_tray_menu(app, app_state.inner()) {
if let Some(tray) = app.tray_by_id("main") {
if let Err(e) = tray.set_menu(Some(new_menu)) {
@@ -142,6 +130,6 @@ impl FailoverSwitchManager {
}
}
Ok(true)
Ok(switched)
}
}
+1
View File
@@ -24,6 +24,7 @@ pub mod response_processor;
pub(crate) mod server;
pub mod session;
pub(crate) mod sse;
pub(crate) mod switch_lock;
pub mod thinking_budget_rectifier;
pub mod thinking_optimizer;
pub mod thinking_rectifier;
+42
View File
@@ -0,0 +1,42 @@
//! Per-app switch lock
//!
//! 确保同一应用同时只有一个供应商切换操作在执行,
//! 防止并发切换导致 is_current 与 Live 备份不一致。
use std::collections::HashMap;
use std::sync::Arc;
use tokio::sync::{Mutex, OwnedMutexGuard, RwLock};
/// 每个应用类型一把互斥锁,保证同一应用的切换操作串行执行。
///
/// 不同应用之间(如 Claude 和 Codex)可以并行切换。
#[derive(Clone, Default)]
pub struct SwitchLockManager {
locks: Arc<RwLock<HashMap<String, Arc<Mutex<()>>>>>,
}
impl SwitchLockManager {
pub fn new() -> Self {
Self::default()
}
/// 获取指定应用的切换锁。
///
/// 返回 `OwnedMutexGuard`,持有期间同一 `app_type` 的其他切换会排队等待。
pub async fn lock_for_app(&self, app_type: &str) -> OwnedMutexGuard<()> {
let lock = {
let locks = self.locks.read().await;
if let Some(lock) = locks.get(app_type) {
lock.clone()
} else {
drop(locks);
let mut locks = self.locks.write().await;
locks
.entry(app_type.to_string())
.or_insert_with(|| Arc::new(Mutex::new(())))
.clone()
}
};
lock.lock_owned().await
}
}