[kernel] Prevent duplicate defunct enqueue for deleted threads.

[Rbb666]

拉取/合并请求描述：(PR description)

为什么提交这份 PR (why to submit this PR)

线程自然退出、显式删除或分离后，在对象仍有效但尚未完成回收时再次调用 rt_thread_delete() / rt_thread_detach()，可能将同一线程重复加入 defunct 队列，造成链表破坏或重复清理。本 PR 为清理过程增加一次性保护。

Prevent duplicate defunct enqueue and cleanup while the thread object is still valid.

你的解决方案是什么 (what is your solution)

• 在 struct rt_thread 中增加独立的 rt_uint8_t defunct 字段，由公共初始化路径 _thread_init() 清零，覆盖动态创建、静态初始化及合法的重新初始化。
• _thread_mark_defunct() 在 thread->spinlock 下完成检查和置位。标志表示“已取得清理权”，在 mutex 清理和入队之前设置，一次生命周期内不再清零。
• 标志使用独立字节，与调度器位域分离，避免不同锁保护下对共享位域存储进行读改写时相互覆盖。注释说明了保护规则和生命周期。
• 保持 close → mark → mutex 清理 → defunct 入队的顺序。线程锁仅保护检查和置位；成功取得清理权的路径执行一次清理和入队，后续重复 delete/detach 返回 -RT_ERROR。
• 保护要求线程对象仍有效，不保证已释放句柄的安全，也不扩展为解决运行中线程的跨核提前回收问题。

Use a separate byte in the thread object, protected by the thread spinlock, to claim cleanup once per lifetime. This avoids sharing storage with scheduler bit-fields that use different locking.

修改文件 / Changed files

• include/rtdef.h：增加独立的线程清理标志。
• src/thread.c：初始化标志，并在退出、删除和分离路径中保护一次性清理。
• src/utest/thread_tc.c：补充重复操作、静态对象复用和双核并发删除回归测试。

测试覆盖 / Regression coverage

• 动态线程自然退出后，在 cleanup 中再次 delete，验证返回 -RT_ERROR 且只清理一次。
• 动态线程显式删除后，在 cleanup 中再次 delete，验证相同的负返回与清理次数。
• 单核、非 SMART 配置下，验证静态线程重复 detach、重新初始化同一 TCB 后自然退出再 detach；通过后续 cleanup 确认旧生命周期已完成回收。
• SMP 至少两核时，两个 worker 分别绑定 CPU 0/1，并发删除同一个未启动的动态线程。目标 cleanup 保持对象存活，直到在途删除完成；验证一次 RT_EOK、一次 -RT_ERROR 和一次 cleanup。
• 测试增加有界等待与超时状态保护。内核调用本身永久卡死仍需外部 watchdog 判失败。

请提供验证的 BSP 和 config (provide the config and BSP)

• BSP：bsp/qemu-vexpress-a9，分别使用 UP（1 核）和 SMP（2 核）配置，均未启用 SMART。
• 工具：Arm GNU Toolchain 13.3.1、SCons 4.10.1、QEMU ARM 8.0.94。
• 验证日期：2026-10-17。
• 在临时 BSP 副本中从 .config 生成 rtconfig.h，仓库 BSP 配置未改动。

两种配置均在默认 BSP 配置上启用：

CONFIG_RT_USING_UTESTCASES=y
CONFIG_RT_UTEST_THREAD=y
CONFIG_UTEST_THR_STACK_SIZE=8192

SMP 配置额外启用：

CONFIG_RT_USING_SMP=y
CONFIG_RT_CPUS_NR=2
CONFIG_SYSTEM_THREAD_STACK_SIZE=4096

构建与运行命令：

python -m SCons --pyconfig-silent
python -m SCons -j8 --strict
utest_run -thread core.thread_tc

QEMU 使用 -M vexpress-a9，UP 为 -smp 1，SMP 为 -smp 2。

配置	严格编译	core.thread_tc
Cortex-A9 UP / 1 核	通过	通过，运行 1 次
Cortex-A9 SMP / 2 核	通过	通过，运行 1 次

两次运行均输出：

[  PASSED  ] [ result   ] testcase (core.thread_tc)
[==========] [ utest    ] finished

git diff --check 通过。在上述配置中，sizeof(struct rt_thread) 分别由 UP 的 168 增至 172 字节、SMP 的 176 增至 180 字节；需重编依赖线程结构体布局的代码。

验证范围 / Validation limits

上述结果是本地严格构建及 QEMU 单次回归结果，不代表全部 utest 或压力测试通过。AArch64 / ARCH_USING_HW_THREAD_SELF 的运行路径尚未实测；SMP 新增用例的目标未启动，不宣称覆盖运行中线程自然退出与远端删除的专项竞态。

Both local UP and SMP strict builds and core.thread_tc runs passed. AArch64 runtime behavior and the running-thread exit-versus-delete race were not tested.

• action：本地验证结果如上；本次推送后的 GitHub Actions 结果以 PR Checks 为准。

当前拉取/合并请求的状态 Intent for your PR

必须选择一项 Choose one (Mandatory):

• 本拉取/合并请求是一个草稿版本 This PR is for a code-review and is intended to get feedback
• 本拉取/合并请求是一个成熟版本 This PR is mature, and ready to be integrated into the repo

代码质量 Code Quality：

我在这个拉取/合并请求中已经考虑了 As part of this pull request, I've considered the following:

• 已经仔细查看过代码改动的对比 Already check the difference between PR and old code
• 代码风格正确，包括缩进空格，命名及其他风格 Style guide is adhered to, including spacing, naming and other styles
• 没有垃圾代码，代码尽量精简，不包含#if 0代码，不包含已经被注释了的代码 All redundant code is removed and cleaned up
• 所有变更均有原因及合理的，并且不会影响到其他软件组件代码或BSP All modifications are justified and not affect other components or BSP
• 对难懂代码均提供对应的注释 I've commented appropriately where code is tricky
• 代码是高质量的 Code in this PR is of high quality
• 已经使用formatting 等源码格式化工具确保格式符合RT-Thread代码规范 This PR complies with RT-Thread code specification
• 如果是新增bsp, 已经添加ci检查到.github/ALL_BSP_COMPILE.json 详细请参考链接BSP自查

[github-actions]

👋 感谢您对 RT-Thread 的贡献！Thank you for your contribution to RT-Thread!

为确保代码符合 RT-Thread 的编码规范，请在你的仓库中执行以下步骤运行代码格式化工作流（如果格式化CI运行失败）。
To ensure your code complies with RT-Thread's coding style, please run the code formatting workflow by following the steps below (If the formatting of CI fails to run).

---

🛠 操作步骤 | Steps

1. 前往 Actions 页面 | Go to the Actions page
   点击进入工作流 → | Click to open workflow →

2. 点击 Run workflow | Click Run workflow

• Use workflow from 保持默认分支（通常为 master）
  Keep the default branch (usually master) in Use workflow from
• 在 branch 输入框填写 PR 分支 defunct
  Enter PR branch defunct in the branch field
• 设置需排除的文件/目录（目录请以"/"结尾）
  Set files/directories to exclude (directories should end with "/")

3. 等待工作流完成 | Wait for the workflow to complete
   格式化后的代码将作为独立提交推送至你的分支。
   The formatting changes will be pushed to your branch as a separate commit.

完成后，提交将自动更新至 defunct 分支，关联的 Pull Request 也会同步更新。
Once completed, commits will be pushed to the defunct branch automatically, and the related Pull Request will be updated.

如有问题欢迎联系我们，再次感谢您的贡献！💐
If you have any questions, feel free to reach out. Thanks again for your contribution!

[github-actions]

📌 Code Review Assignment

🏷️ Tag: kernel

Reviewers: @GorrayLi @ReviewSun @hamburger-os @lianux-mm @wdfk-prog @xu18838022837

Changed Files (Click to expand)

• src/thread.c
• src/utest/thread_tc.c

---

📊 Current Review Status (Last Updated: 2026-10-17 16:04 CST)

• ⌛ @GorrayLi Pending Review
• ⌛ @ReviewSun Pending Review
• ⌛ @hamburger-os Pending Review
• ⌛ @lianux-mm Pending Review
• ⌛ @wdfk-prog Pending Review
• ⌛ @xu18838022837 Pending Review

---

📝 Review Instructions

1. 维护者可以通过单击此处来刷新审查状态: 🔄 刷新状态
   Maintainers can refresh the review status by clicking here: 🔄 Refresh Status

2. 确认审核通过后评论 LGTM/lgtm
   Comment LGTM/lgtm after confirming approval

3. PR合并前需至少一位维护者确认
   PR must be confirmed by at least one maintainer before merging

ℹ️ 刷新CI状态操作需要具备仓库写入权限。
ℹ️ Refresh CI status operation requires repository Write permission.

[github-actions]

MemBrowse Memory Report

at32f415-start

• ROM: .text +80 B (+0.1%, 65,500 B / 262,144 B, total: 25% used)

esp32-c3

• drom0_0_seg: .eh_frame +40 B, .eh_frame_hdr +8 B (+0.0%, 411,036 B / 8,388,576 B, total: 5% used)

gd32105r-start

• CODE: .text +80 B (+0.1%, 69,416 B / 262,144 B, total: 26% used)

hc32f334

• FLASH: .text +88 B (+0.1%, 99,680 B / 131,072 B, total: 76% used)

hpmicro-hpm5301evklite

• DLM: .bss +16 B (+0.0%, 58,272 B / 130,304 B, total: 45% used)
• ILM: .fast +144 B (+0.4%, 38,384 B / 131,072 B, total: 29% used)
• XPI0: .fast +144 B (+0.1%, 125,144 B / 1,048,576 B, total: 12% used)

infineon-psoc6

• flash: .text +76 B (+0.1%, 116,268 B / 262,144 B, total: 44% used)

k230

• SRAM: .bss +32 B, .eh_frame +72 B (+0.0%, 1,128,643 B / 268,300,288 B, total: 0% used)

loongson-ls1cdev

• Code: .text +184 B (+0.1%, 363,544 B)
• Data: .bss +16 B (+0.0%, 59,828 B)

nordic-nrf51822

• FLASH: .text +116 B (+0.2%, 52,152 B / 262,144 B, total: 20% used)
• RAM: .bss +4 B (+0.1%, 2,720 B / 16,384 B, total: 17% used)

nuvoton-m487

• CODE: .text +104 B (+0.0%, 356,280 B / 524,288 B, total: 68% used)
• DATA: .bss +24 B (+0.0%, 101,956 B / 163,840 B, total: 62% used)

qemu-virt64-aarch64

• Code: .eh_frame +32 B, .text +192 B (+0.0%, 1,078,036 B)

raspberry-pico-rp2040

• FLASH: .text +104 B (+0.1%, 114,272 B / 2,097,152 B, total: 5% used)
• RAM: .bss +4 B (+0.1%, 5,712 B / 262,144 B, total: 2% used)

renesas-ra2l1

• FLASH: .text +76 B (+0.1%, 98,352 B / 262,144 B, total: 38% used)
• RAM: .bss +8 B (+0.1%, 6,648 B / 32,768 B, total: 20% used)

simulator

• Code: .eh_frame +32 B, .eh_frame_hdr +8 B, .rela.dyn +1,080 B, .rodata +192 B, .text +766 B (+0.1%, 2,675,098 B)
• Data: .bss +32 B, .data +1,280 B (+0.1%, 1,610,596 B)

stm32f407-rt-spark

• CODE: .text +76 B (+0.1%, 86,144 B / 1,048,576 B, total: 8% used)

stm32l475-atk-pandora-llvm

• RAM1: .bss +8 B (+0.2%, 4,252 B / 98,304 B, total: 4% used)
• ROM: .text +72 B (+0.1%, 82,760 B / 524,288 B, total: 16% used)

wch-ch32v208w-r0

• FLASH: .text +136 B (+0.1%, 159,736 B / 491,520 B, total: 32% used)
• RAM: .bss +8 B (+0.0%, 47,952 B / 65,536 B, total: 73% used)

x86

• Code: .text +128 B (+0.1%, 199,053 B)
• Data: .bss +64 B (+0.4%, 14,644 B)

xuantie-e901plus

• DSRAM: .bss +4 B (+0.0%, 14,016 B / 524,288 B, total: 3% used)
• ISRAM: .rodata +16 B, .text +48 B (+0.1%, 46,424 B / 131,072 B, total: 35% used)
  No memory changes detected for:
• nxp-lpc1114

[BernardXiong]

Review / 审查

结论 / Conclusion: no blocking issue found in the duplicate-defunct protection.

The new flag is initialized with the scheduler context, and _thread_mark_defunct() makes the check-and-set atomic under the thread spinlock before mutex detachment and defunct enqueue. This addresses the duplicate enqueue/double cleanup scenario described by the PR, and the change is small and localized.

建议 / Suggestions

1. Please add an SMP-focused regression case, or document why the existing lock ordering is sufficient. The paths now combine the scheduler critical section, the thread spinlock, and mutex detachment; a test should exercise concurrent rt_thread_delete()/rt_thread_detach() against natural thread exit.

2. The new bit lives in rt_sched_thread_ctx, whose other fields are generally accessed under scheduler-lock rules, while _thread_mark_defunct() uses the thread spinlock. Please add a short comment documenting that sched_flag_defunct is intentionally protected by thread->spinlock, so future scheduler changes do not introduce a lock-domain mismatch.

3. Keep the negative return behavior covered for both dynamic and static threads after natural exit, including repeated calls. The current tests cover the main cases; the SMP/race case is the remaining useful gap.

The PR description and test intent are clear. / PR 描述和测试意图清晰。

[Rbb666]

@BernardXiong 熊大，已经按照修改建议修复了