Add rt_scheduler_critical_switch_flag to reduce scheduling on UP and fix bug on rzn2l #10581

PillarShi · 2025-08-04T13:47:56Z

拉取/合并请求描述：(PR description)

[

为什么提交这份PR (why to submit this PR)

rzn2l 中断重复调用rt_interrupt_enter以及rt_interrupt_leave
rzn2l定时器驱动在默认直接编译运行是有问题的
rzn2l无法开启高优化，工具生成的代码有问题
优化UP下的性能

你的解决方案是什么 (what is your solution)

删除重复
更改驱动
加volatile
加入rt_scheduler_critical_switch_flag，使单核和多核策略同步

更改前

更改后

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

BSP: bsp\renesas\rzn2l_etherkit

]

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/workflows/bsp_buildings.yml 详细请参考链接BSP自查

CLAassistant · 2025-08-04T13:48:02Z

All committers have signed the CLA.

github-actions · 2025-08-04T13:48:06Z

📌 Code Review Assignment

🏷️ Tag: bsp_renesas

Reviewers: kurisaW

Changed Files (Click to expand)

bsp/renesas/libraries/HAL_Drivers/config/rzt/timer_config.h
bsp/renesas/libraries/HAL_Drivers/drv_hwtimer.c
bsp/renesas/rzn2l_etherkit/rzn/fsp/src/bsp/cmsis/Device/RENESAS/Source/system.c

🏷️ Tag: kernel

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

Changed Files (Click to expand)

src/cpu_up.c
src/scheduler_up.c

📊 Current Review Status (Last Updated: 2025-08-06 15:24 CST)

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

📝 Review Instructions

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

Rbb666 · 2025-08-04T15:20:58Z

src/scheduler_up.c

        rt_hw_interrupt_enable(level);

-        if (rt_current_thread)
+        if (rt_scheduler_critical_switch_flag == 1)


rt_schedule里面开始就会进行是否锁调度器的判断，没看出来这么修改后如何提升性能的？
麻烦举例说明下，谢谢

更新：这个位置的修改是为了减少调度的次数吧？

是的，针对重复的调度进入，这是改动最小的方法吧，更改前后的对比可以看描述中的图（-O0时的数据）

Rbb666 · 2025-08-05T01:49:23Z

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.You have signed the CLA already but the status is still pending? Let us recheck it.

麻烦签署下CLA，PR的标题也修改下吧，可以说明清晰点解决了什么问题

Copilot

Pull Request Overview

This PR adds a critical switch flag mechanism to optimize scheduler performance on single-core systems and fixes multiple issues in the rzn2l BSP, including duplicate interrupt handling and timer driver problems.

Introduces rt_scheduler_critical_switch_flag to track deferred scheduling needs and optimize performance on UP systems
Removes duplicate rt_interrupt_enter/leave calls in rzn2l timer callbacks that were causing issues
Fixes timer configuration and adds volatile keyword to prevent optimization issues

Reviewed Changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
src/scheduler_up.c	Adds critical switch flag mechanism to defer scheduling until safe to execute
src/cpu_up.c	Updates documentation to reflect deferred scheduling behavior
bsp/renesas/rzn2l_etherkit/rzn/fsp/src/bsp/cmsis/Device/RENESAS/Source/system.c	Adds volatile keyword to counter variables to prevent optimization issues
bsp/renesas/libraries/HAL_Drivers/drv_hwtimer.c	Removes duplicate interrupt handling for rzn2l and adds missing prescaler values
bsp/renesas/libraries/HAL_Drivers/config/rzt/timer_config.h	Increases prescaler array size to accommodate new values

Comments suppressed due to low confidence (2)

src/scheduler_up.c:53

[nitpick] The variable name rt_scheduler_critical_switch_flag is overly verbose. Consider a shorter, clearer name like rt_sched_deferred_flag or rt_critical_sched_flag.

static rt_int8_t rt_scheduler_critical_switch_flag = -1;

bsp/renesas/libraries/HAL_Drivers/drv_hwtimer.c:22

The array size increase from 8 to 9 in timer_config.h doesn't match the addition of only 2 new prescaler values (lines 41-42). This could lead to array bounds issues or uninitialized elements.

#ifdef BSP_USING_TIM0

Copilot · 2025-08-05T10:54:36Z

src/scheduler_up.c


 extern volatile rt_atomic_t rt_interrupt_nest;
 static rt_int16_t rt_scheduler_lock_nest;
+static rt_int8_t rt_scheduler_critical_switch_flag = -1;


The magic number -1 for initialization is unclear. Consider using a named constant or enum to make the different states more explicit (e.g., SCHED_FLAG_UNINITIALIZED = -1, SCHED_FLAG_IDLE = 0, SCHED_FLAG_PENDING = 1).

Suggested change

static rt_int8_t rt_scheduler_critical_switch_flag = -1;

static rt_int8_t rt_scheduler_critical_switch_flag = SCHED_FLAG_UNINITIALIZED;

Copilot · 2025-08-05T10:54:37Z

src/scheduler_up.c

        rt_hw_interrupt_enable(level);

-        if (rt_current_thread)
+        if (rt_scheduler_critical_switch_flag == 1)


The magic number 1 should be replaced with a named constant to improve code clarity and maintainability.

Suggested change

if (rt_scheduler_critical_switch_flag == 1)

if (rt_scheduler_critical_switch_flag == RT_SCHEDULER_CRITICAL_SWITCH_PENDING)

PillarShi · 2025-08-05T11:51:53Z

AI审查真棒啊，那需要这样改吗，MP那里也是单纯的使用数字，两边同步个宏？

我想的是减少影响范围和差异，因此也直接使用数字

Rbb666 · 2025-08-05T12:01:31Z

AI审查真棒啊，那需要这样改吗，MP那里也是单纯的使用数字，两边同步个宏？

我想的是减少影响范围和差异，因此也直接使用数字

感觉宏好一些

ReviewSun · 2025-08-06T00:44:23Z

请修改数字标志位为，方便管理、裁剪。

ReviewSun · 2025-08-06T00:49:58Z

麻烦将commit修改为1个

PillarShi · 2025-08-06T02:25:07Z

更新：

添加标志位宏
合并为一个commit

kurisaW · 2025-08-06T05:18:26Z

bsp/renesas/libraries/HAL_Drivers/drv_hwtimer.c

 #ifdef BSP_USING_TIM0
 void timer0_callback(timer_callback_args_t *p_args)
 {
+#if !defined(SOC_SERIES_R9A07G0)


我觉得如果你想在timer中断回调里去掉RTT的中断嵌套计数，以此来提升程序的性能，在自己的项目中应用就好了，但是这部分代码就不要提到主线了，而且你说的“重复调用rt_interrupt_enter以及rt_interrupt_leave”，没理解你意思

是因为在libc里面会隐式调用一次:

rt-thread/libcpu/arm/cortex-r52/context_gcc.S

Line 149 in 236e830

BL rt_interrupt_enter

那是不是建议直接修改:
rt-thread/libcpu/arm/cortex-r52/context_gcc.S

那上下文切换汇编里就不考虑此操作了，与其他libcpu统一一下规范，这么看也挺别扭的

Rbb666 · 2025-08-05T01:35:49Z

src/scheduler_up.c


 extern volatile rt_atomic_t rt_interrupt_nest;
 static rt_int16_t rt_scheduler_lock_nest;
+static rt_int8_t rt_scheduler_critical_switch_flag = -1;


这块推荐这种写法，尽量不要使用魔法数字，语义不直观：

rt-thread/src/scheduler_mp.c

Line 104 in 32e93ad

#define IS_CRITICAL_SWITCH_PEND(pcpu, curthr) (RT_SCHED_CTX(curthr).critical_switch_flag)

Rbb666 · 2025-08-06T05:21:17Z

bsp/renesas/libraries/HAL_Drivers/drv_hwtimer.c

 #ifdef BSP_USING_TIM0
 void timer0_callback(timer_callback_args_t *p_args)
 {
+#if !defined(SOC_SERIES_R9A07G0)


是因为在libc里面会隐式调用一次:

rt-thread/libcpu/arm/cortex-r52/context_gcc.S

Line 149 in 236e830

BL rt_interrupt_enter

Rbb666 · 2025-08-06T05:22:43Z

麻烦将commit修改为1个

多个commit也没啥问题，只要不冲突&每个commit都说明提交的内容即可

Rbb666 · 2025-08-06T05:41:15Z

那是不是建议直接修改: rt-thread/libcpu/arm/cortex-r52/context_gcc.S

那上下文切换汇编里就不考虑此操作了，与其他libcpu统一一下规范，这么看也挺别扭的

感觉问题不大，rv里面也是这么干的，rt_interrupt_enter 作用是为了让内核判断是否在中断中。
@PillarShi 问下作者，这个PR的主要优化点是在加入rt_scheduler_critical_switch_flag避免 rt_schedule 多次调用吧？

PillarShi · 2025-08-06T06:09:21Z

那我归总一下吧。当前pr的关键点：

避免 rt_schedule 多次调用
修改定时器驱动上的bug

这两个明显影响使用，至于rt_interrupt_enter以及rt_interrupt_leave重复不在这个pr或者主线上更改。

我一会再调整一下：

2个关键点还是分开commit，方便以后review
调整魔数，用更直观的宏
rt_interrupt_enter以及rt_interrupt_leave先还原不改动

amos-su · 2025-11-20T06:45:53Z

这个修改直接导致我线程切换异常了，我恢复成5.2.1版本后，线程切换才变正常

amos-su · 2025-11-20T06:48:08Z

除了这里跟5.2.1版本保持一致，其它代码跟5.2.2版本一模一样，线程切换就正常了

amos-su · 2025-11-20T06:49:38Z

如果使用5.2.2版本，工程跑起来会有很多应用线程跑不起来

Rbb666 · 2025-11-20T06:54:33Z

如果使用5.2.2版本，工程跑起来会有很多应用线程跑不起来

主线暂时没发现这个问题呢，可以给下复现代码

amos-su · 2025-11-20T06:58:37Z

应用代码不好给出，我用的是GD32H757

amos-su · 2025-11-20T07:04:14Z

同样的应用，5.2.2跑起来会经常发生部分应用线程跑不起来(10次有3-4次吧)，然后我是一部分代码一部分代码屏蔽，测试，整整搞了一周，才发现是这里导致的

amos-su · 2025-11-20T08:06:03Z

在应用线程初始化阶段,有时候会发生某些线程启动失败且某些线程没有初始化的现象,应该是线程切换出现了问题

PillarShi · 2025-12-08T01:59:34Z

在应用线程初始化阶段,有时候会发生某些线程启动失败且某些线程没有初始化的现象,应该是线程切换出现了问题

解决了吗？看这句话的现象，如果初始化都没好，那肯定无法完美启动，更不用说调度了。

PillarShi requested a review from Rbb666 as a code owner August 4, 2025 13:47

github-actions bot added BSP Kernel PR has src relate code BSP: Renesas BSP related with Renesas labels Aug 4, 2025

Rbb666 reviewed Aug 4, 2025

View reviewed changes

PillarShi changed the title ~~Pillar rzn2l~~ Add rt_scheduler_critical_switch_flag to reduce scheduling on UP and fix bug on rzn2l Aug 5, 2025

Rbb666 requested a review from Copilot August 5, 2025 10:53

Copilot AI reviewed Aug 5, 2025

View reviewed changes

PillarShi force-pushed the pillar-rzn2l branch from df99e58 to 8c915e1 Compare August 6, 2025 02:22

kurisaW reviewed Aug 6, 2025

View reviewed changes

Rbb666 reviewed Aug 6, 2025

View reviewed changes

PillarShi added 2 commits August 6, 2025 14:22

[bsp][etherkit] fix bug and enable high optimization.

9636b3c

[kernel] add UP scheduler critical switch flag.

ec9f176

PillarShi force-pushed the pillar-rzn2l branch from 8c915e1 to ec9f176 Compare August 6, 2025 07:23

Rbb666 approved these changes Aug 6, 2025

View reviewed changes

Rbb666 merged commit b084503 into RT-Thread:master Aug 6, 2025
3 checks passed

Rbb666 added this to the v5.2.2 milestone Sep 23, 2025

PillarShi deleted the pillar-rzn2l branch December 8, 2025 01:49

	static rt_int8_t rt_scheduler_critical_switch_flag = -1;
	static rt_int8_t rt_scheduler_critical_switch_flag = SCHED_FLAG_UNINITIALIZED;

	if (rt_scheduler_critical_switch_flag == 1)
	if (rt_scheduler_critical_switch_flag == RT_SCHEDULER_CRITICAL_SWITCH_PENDING)

Add rt_scheduler_critical_switch_flag to reduce scheduling on UP and fix bug on rzn2l #10581

Add rt_scheduler_critical_switch_flag to reduce scheduling on UP and fix bug on rzn2l #10581

Conversation

PillarShi commented Aug 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

拉取/合并请求描述：(PR description)

为什么提交这份PR (why to submit this PR)

你的解决方案是什么 (what is your solution)

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

当前拉取/合并请求的状态 Intent for your PR

代码质量 Code Quality：

Uh oh!

CLAassistant commented Aug 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions bot commented Aug 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

📌 Code Review Assignment

🏷️ Tag: bsp_renesas

🏷️ Tag: kernel

📊 Current Review Status (Last Updated: 2025-08-06 15:24 CST)

📝 Review Instructions

Uh oh!

Rbb666 Aug 4, 2025

Choose a reason for hiding this comment

Uh oh!

Rbb666 Aug 4, 2025

Choose a reason for hiding this comment

Uh oh!

PillarShi Aug 5, 2025

Choose a reason for hiding this comment

Uh oh!

Rbb666 commented Aug 5, 2025

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

Copilot AI Aug 5, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Aug 5, 2025

Choose a reason for hiding this comment

Uh oh!

PillarShi commented Aug 5, 2025

Uh oh!

Rbb666 commented Aug 5, 2025

Uh oh!

ReviewSun commented Aug 6, 2025

Uh oh!

ReviewSun commented Aug 6, 2025

Uh oh!

PillarShi commented Aug 6, 2025

Uh oh!

kurisaW Aug 6, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Rbb666 Aug 6, 2025

Choose a reason for hiding this comment

Uh oh!

kurisaW Aug 6, 2025

Choose a reason for hiding this comment

Uh oh!

Rbb666 Aug 5, 2025

Choose a reason for hiding this comment

Uh oh!

Rbb666 Aug 6, 2025

Choose a reason for hiding this comment

Uh oh!

Rbb666 commented Aug 6, 2025

Uh oh!

Rbb666 commented Aug 6, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

PillarShi commented Aug 6, 2025

Uh oh!

Uh oh!

amos-su commented Nov 20, 2025

PillarShi commented Aug 4, 2025 •

edited

Loading

CLAassistant commented Aug 4, 2025 •

edited

Loading

github-actions bot commented Aug 4, 2025 •

edited

Loading

kurisaW Aug 6, 2025 •

edited

Loading

Rbb666 commented Aug 6, 2025 •

edited

Loading