[Interrupt reasoning] Add interrupt_requests control command support

[lonelygsh]

Motivation

支持接受中断推理的命令。

Modifications

在 internal_adapter_utils.py的 handle_control_cmd 方法中，新增 interrupt_requests 控制命令的处理分支：

接收 interrupt_requests 命令后，从 task["req_ids"] 中提取需要中断的请求 ID 列表
调用 self.engine.resource_manager.add_abort_req_ids() 将这些请求标记为待中止
构造包含 success 状态和已中断请求 ID 列表的响应，通过控制命令通道回传结果

Usage or Command

{
"cmd": "interrupt_requests",
"task_id": "<task_id>",
"req_ids": ["req_1", "req_2"]
}

Accuracy Tests

本次改动不涉及模型前向计算或 kernel 修改，无需精度测试。

Checklist

• Add at least a tag in the PR title.
  • Tag list: [[FDConfig],[APIServer],[Engine], [Scheduler], [PD Disaggregation], [Executor], [Graph Optimization], [Speculative Decoding], [RL], [Models], [Quantization], [Loader], [OP], [KVCache], [DataProcessor], [BugFix], [Docs], [CI], [Optimization], [Feature], [Benchmark], [Others], [XPU], [HPU], [GCU], [DCU], [Iluvatar], [Metax]]
  • You can add new tags based on the PR content, but the semantics must be clear.
• Format your code, run pre-commit before commit.
• Add unit tests. Please write the reason in this PR if no unit tests.
• Provide accuracy results.不涉及精度变化
• If the current PR is submitting to the release branch, make sure the PR has been submitted to the develop branch, then cherry-pick it to the release branch with the [Cherry-Pick] PR tag.

[paddle-bot]

Thanks for your contribution!

[PaddlePaddle-bot]

🤖 AI Code Review | 2026-04-16 21:15 CST\n\n## 📋 Review 摘要\n\nPR 概述：在 internal_adapter_utils.py 中新增 interrupt_requests 控制命令，支持通过控制通道中断指定请求的推理。\n变更范围：fastdeploy/splitwise/internal_adapter_utils.py\n影响面 Tag：Engine\n\n### 📝 PR 规范检查\n\nPR 标题中的 Tag [Interrupt reasoning] 不在官方 Tag 列表中，建议使用官方标签。\n\n标题建议（可直接复制）：\n- [Feature] Add interrupt_requests control command support\n\n### 问题\n\n| 级别 | 文件 | 概述 |\n|------|------|------|\n| 🟡 建议 | internal_adapter_utils.py:109 | 缺少对 req_ids 字段的防御性校验，异常时调用方无法收到响应 |\n| 🟡 建议 | internal_adapter_utils.py:108 | 缺少 logger.debug 日志，与其他命令分支风格不一致 |\n\n### 总体评价\n\n变更逻辑清晰，整体遵循了现有控制命令的处理模式。建议补充防御性校验和调试日志以提升健壮性和可观测性。"

[PaddlePaddle-bot]

🟡 建议 缺少对 req_ids 字段的防御性校验。

如果传入的 task 缺少 req_ids 键或其值为空列表，当前代码会抛出 KeyError 或执行一次无意义的空调用。虽然外层 except Exception 会捕获异常不会导致线程崩溃，但调用方不会收到任何响应，可能导致请求超时。

建议添加校验：

elif task["cmd"] == "interrupt_requests":
    req_ids = task.get("req_ids", [])
    if not req_ids:
        result = {
            "task_id": task_id_str,
            "result": {"success": False, "message": "req_ids is empty or missing"},
        }
    else:
        self.engine.resource_manager.add_abort_req_ids(req_ids)
        result = {
            "task_id": task_id_str,
            "result": {"success": True, "interrupted_req_ids": req_ids},
        }
    with self.response_lock:
        self.recv_control_cmd_server.response_for_control_cmd(task_id_str, result)

[PaddlePaddle-bot]

🟡 建议 其他命令分支（如 get_payload、check_health）在响应前都有 logger.debug(f"Response for task: {task_id_str}")，建议此处也添加以保持一致性，方便排查问题。

elif task["cmd"] == "interrupt_requests":
    self.engine.resource_manager.add_abort_req_ids(task["req_ids"])
    result = {
        "task_id": task_id_str,
        "result": {"success": True, "interrupted_req_ids": task["req_ids"]},
    }
    logger.debug(f"Response for task: {task_id_str}")
    with self.response_lock:
        self.recv_control_cmd_server.response_for_control_cmd(task_id_str, result)

[codecov-commenter]

Codecov Report

❌ Patch coverage is 0% with 5 lines in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (develop@f7a2418). Learn more about missing BASE report.

Files with missing lines	Patch %	Lines
fastdeploy/splitwise/internal_adapter_utils.py	0.00%	5 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             develop    #7445   +/-   ##
==========================================
  Coverage           ?   73.89%           
==========================================
  Files              ?      398           
  Lines              ?    54948           
  Branches           ?     8608           
==========================================
  Hits               ?    40603           
  Misses             ?    11630           
  Partials           ?     2715           

Flag	Coverage Δ
GPU	73.89% <0.00%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.