Ascend/msagent / skills/operator/mssanitizer-basic-usage
mssanitizer-basic-usage
Verified昇腾AscendC算子异常检测工具(msSanitizer)。检测内存越界、内存泄漏、数据竞争、未初始化读取、同步异常等缺陷。触发关键词:mssanitizer、sanitizer、检测工具、内存检测、竞争检测、初始化检测、同步检测
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 24.7 KB |
| evals/evals.json | config | 6.8 KB |
| references/alarms.md | markdown | 29.6 KB |
| references/official_workflow.md | markdown | 25.1 KB |
| references/sample_initcheck/CMakeLists.txt | markdown | 1.2 KB |
| references/sample_initcheck/README.md | markdown | 4.8 KB |
| references/sample_initcheck/initcheck.asc | other | 2.3 KB |
| references/sample_memcheck/CMakeLists.txt | markdown | 1.2 KB |
| references/sample_memcheck/README.md | markdown | 4.9 KB |
| references/sample_memcheck/memcheck.asc | other | 2.4 KB |
| references/sample_racecheck/CMakeLists.txt | markdown | 1.2 KB |
| references/sample_racecheck/README.md | markdown | 6.6 KB |
| references/sample_racecheck/racecheck.asc | other | 2.5 KB |
| references/sample_synccheck/CMakeLists.txt | markdown | 1.2 KB |
| references/sample_synccheck/README.md | markdown | 4.1 KB |
| references/sample_synccheck/synccheck.asc | other | 2.8 KB |
| scripts/parse_mssanitizer_report.py | script | 19.0 KB |
| scripts/run_mssanitizer.py | script | 12.3 KB |
references/sample_memcheck/README.md
View on GitHub# msSanitizer 算子内存越界检测样例介绍
## 概述
本样例演示如何使用 **mssanitizer** 工具对算子进行 UB(Unified Buffer)缓冲区越界写入异常的检测和定位。
## 适用场景
- 算子运行时出现异常值、崩溃或结果不稳定,疑似存在越界内存访问。
## 支持的产品范围
- Ascend 950PR/Ascend 950DT
- Atlas A3训练系列产品/Atlas A3推理系列产品
- Atlas A2训练系列产品/Atlas A2推理系列产品
## 目录结构
```text
├── sample_memcheck/
│ ├── CMakeLists.txt // 编译工程文件
│ ├── memcheck.asc // Ascend C 样例实现(含越界写入注入)
│ └── README.md // 本说明文件
```
## 样例描述
### 核心逻辑
本样例在 `memcheck.asc` 中实现了一个基础的搬入(DataCopy)算子。正常流程为:从 Global Memory 搬入数据到 Local Memory(`AscendC::DataCopy(localBuf, srcGlobal, elemCnt)`),搬运量应与 UB 缓冲区分配大小一致。
### 注入异常
在算子中分配了一个仅能容纳 `allocCnt`(64)个 float 元素的 UB 缓冲区,但 `DataCopy` 却尝试搬运 `elemCnt`(256)个元素:
```cpp
// 分配 64 个 float 的 UB 空间(不足)
AscendC::LocalTensor<float> localBuf = ubAlloc.Alloc<float, allocCnt>();
// 试图搬运 256 个 float 到仅 64 个 float 的 UB 缓冲区 → 越界写入!
AscendC::DataCopy(localBuf, srcGlobal, elemCnt);
```
由于 `elemCnt`(256)远大于 UB 实际分配量 `allocCnt`(64),超出部分将写入未分配/他人使用的 UB 空间,触发内存越界告警。
## 编译运行
### 环境准备
请参照官方文档完成开发环境配置:[算子工具开发环境安装指导](https://gitcode.com/Ascend/msot/blob/master/docs/zh/common/dev_env_setup.md)。
### 编译算子
在 `sample_memcheck/` 目录下执行:
```bash
mkdir -p build && cd build
cmake -DCMAKE_ASC_ARCHITECTURES=dav-2201 ..
make -j96
```
编译完成后,`build` 目录下将生成算子二进制文件 `demo`。
> **说明**:编译选项 `--npu-arch` 已通过 CMakeLists.txt 自动配置。若需检测自己的算子,请确保在编译参数中增加 `-g --cce-enable-sanitizer`,可在 CMakeLists.txt 中追加相应编译选项。
- 编译选项说明
| 选项 | 可选值 | 说明 |
|------|--------|------|
| `CMAKE_ASC_RUN_MODE` | `npu`(默认)、`cpu`、`sim` | 运行模式:NPU 运行、CPU调试、NPU仿真 |
| `CMAKE_ASC_ARCHITECTURES` | `dav-2201`(默认)、`dav-3510` | NPU 架构:dav-2201 对应 Atlas A2/A3系列产品,dav-3510 对应 Ascend 950PR/Ascend 950DT |
> **注意:** 切换编译模式前需清理 cmake 缓存,可在 build 目录下执行 `rm CMakeCache.txt` 后重新 cmake。
### 算子运行
1. 拉起算子
使用 mssanitizer 工具,指定 memcheck 模式拉起算子(memcheck 为默认模式,可省略 `-t` 参数):
```bash
mssanitizer -t memcheck ./demo
```
> **说明**:mssanitizer命令参数含义请参考:[mssanitizer 用户指南](https://www.hiascend.com/document/detail/zh/canncommercial/900/devaids/optool/docs/zh/user_guide/mssanitizer_user_guide.md)。
工具成功拉起算子后,将在终端输出如下日志:
```text
[mssanitizer] logging to file: ./mindstudio_sanitizer_log/mssanitizer_XXX.log
[mssanitizer] Start memcheck sanitizer on kernel "bounds_write_demo(unsigned char*, unsigned char*)"
```
2. 检测报告分析
工具扫描完算子后,将输出以下异常日志:
```text
====== ERROR: illegal write of size 768
====== at 0x0 on UB in "bounds_write_demo(unsigned char*, unsigned char*)"
====== in block aiv(0) on device 0
====== code in pc current 0x65c (serialNo:12)
====== #0 /path/to/sample_memcheck/memcheck.asc:33:9
====== #1 /path/to/sample_memcheck/memcheck.asc:47:8
```
768 字节恰好等于 `(256 - 64) × sizeof(float)`,即超出 UB 分配范围的溢出量。根据诊断报告,可以确认 `memcheck.asc` 第 33 行(即 `DataCopy`)向 UB 缓冲区写入了超出其分配范围的数据。修复方式为将 UB 分配大小调整为与实际搬运量一致:
```cpp
// 修复:分配足够大小的 UB 缓冲区
AscendC::LocalTensor<float> localBuf = ubAlloc.Alloc<float, elemCnt>();
AscendC::DataCopy(localBuf, srcGlobal, elemCnt);
```
## 检测流程总结
1. 编译生成 demo 二进制文件
2. 拉起算子 `mssanitizer -t memcheck ./demo`
3. 分析诊断报告,定位源码异常行并修复
## 注意事项
- 检测自定义算子前,请确认编译选项中包含 `-g --cce-enable-sanitizer`,否则无法使用工具检测能力。
- 内存检测为默认检测模式,运行 `mssanitizer ./demo` 等价于 `mssanitizer -t memcheck ./demo`。
- 可通过 `--leak-check=yes` 额外开启内存泄漏检测,`--check-unused-memory=yes` 开启分配未使用检测。
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 193k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 193k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 193k | Updated | Claude CodeCodex | Verified | Install |