verify-bpf

作者: microsoft

為ebpf-for-windows專案建置並驗證eBPF程式。當要求將.c BPF原始碼編譯為.o、使用bpf2c驗證.o ELF檔案時,使用此技能。

npx skills add https://github.com/microsoft/ebpf-for-windows --skill verify-bpf

Build & Verify eBPF Programs (ebpf-for-windows)

Compile eBPF C source to ELF bytecode with clang, then verify (and optionally generate native code) with bpf2c. On verification failure, diagnose using the PREVAIL reference.

When to Use

  • User asks to compile/build a .c eBPF program to .o
  • User asks to verify a .o ELF file or check if it passes the verifier
  • User asks to debug or fix a verification failure from bpf2c output
  • User mentions bpf2c, clang+bpf, or PREVAIL verification in the ebpf-for-windows context

Prerequisites

  • clang.exe — find the first existing path (highest priority first): packages\llvm.tools\clang.exe, "$env:ProgramFiles\LLVM\bin\clang.exe", "$(& "${env:ProgramFiles(x86)}\Microsoft Visual Studio\Installer\vswhere.exe" -latest -property installationPath)\VC\Tools\Llvm\bin\clang.exe"
  • bpf2c.exe — if x64\Debug\bpf2c.exe is missing, build it: msbuild ebpf-for-windows.sln /m /p:Configuration=Debug /p:Platform=x64 /t:tools\bpf2c

Quick Reference

1. Compile: C → ELF (.o)

Run from the solution root directory (use the clang path found above):

# Standard sample programs
& '<clang-path>' -g -target bpf -O2 -Werror `
  -Iinclude -Iexternal\bpftool `
  -Itests\xdp -Itests\socket `
  -Itests\sample\ext\inc -Itests\include `
  -c <SOURCE>.c -o <OUTPUT>.o

For undocked programs (tests\sample\undocked\*.c), also add:

  • -Itests\sample
  • -Iundocked\tests\sample\ext\inc

2. Verify: ELF (.o) → native C (via bpf2c)

.\x64\Debug\bpf2c.exe --bpf <FILE>.o --sys <OUTPUT_DIR>\<name>_driver.c

Add --verbose to get detailed verifier output (pre/post invariants at each instruction) on failure.

bpf2c Options

FlagPurpose
--bpf <file>Input ELF file containing BPF bytecode
--sys <file>Generate Windows kernel driver C wrapper
--dll <file>Generate Windows DLL C wrapper
--raw <file>Generate C code without platform wrapper
--verboseShow detailed verifier failure info (invariants)
--type <str>Override eBPF program type string
--hash <alg>Algorithm used to hash ELF file

Workflow

Step 1: Identify the Source

Determine the .c source file and where the .o should go. Common patterns:

Source LocationInclude Flags
tests\sample\*.c-Iinclude -Iexternal\bpftool -Itests\xdp -Itests\socket -Itests\sample\ext\inc -Itests\include
tests\sample\undocked\*.cSame as above, plus -Itests\sample -Iundocked\tests\sample\ext\inc
User-provided fileStart with the standard sample flags; add more -I paths as needed

Step 2: Compile with Clang

Run clang. If it fails, fix compiler errors in the C source (standard C/clang diagnostics).

Step 3: Verify with bpf2c

Run bpf2c with --sys (or --dll/--raw). Two outcomes:

  • Success → bpf2c generates the output C file. Verification passed.
  • Failure → bpf2c prints verifier errors. Proceed to diagnosis.

Step 4: Diagnose Failures

On verification failure:

  1. Re-run with --verbose if not already used, to get full invariant output.
  2. Read external/ebpf-verifier/docs/llm-context.md — the authoritative PREVAIL diagnostic reference.
  3. Follow the diagnosis protocol in that document (identify error → check pre-invariant → trace root cause → recommend fix).

Example: Full Pipeline

# Compile (undocked program — needs full include paths)
& '<clang-path>' -g -target bpf -O2 -Werror `
  -Iinclude -Iexternal\bpftool `
  -Itests\xdp -Itests\socket -Itests\sample\ext\inc -Itests\include `
  -Itests\sample -Iundocked\tests\sample\ext\inc `
  -c tests\sample\undocked\perf_event_burst.c `
  -o x64\Debug\perf_event_burst.o

# Verify + generate driver
.\x64\Debug\bpf2c.exe --bpf x64\Debug\perf_event_burst.o `
  --sys .\x64\Debug\perf_event_burst_km\perf_event_burst_driver.c

# If verification fails, re-run with --verbose for diagnosis
.\x64\Debug\bpf2c.exe --bpf x64\Debug\perf_event_burst.o `
  --sys .\x64\Debug\perf_event_burst_km\perf_event_burst_driver.c `
  --verbose

來自 microsoft 的更多技能

oss-growth
microsoft
開源增長駭客角色
agent-framework-azure-ai-py
microsoft
使用Microsoft Agent Framework Python SDK(agent-framework-azure-ai)构建Azure AI Foundry代理。适用于使用AzureAIAgentsProvider创建持久化代理、使用托管工具(代码解释器、文件搜索、网络搜索)、集成MCP服务器、管理对话线程或实现流式响应。涵盖函数工具、结构化输出和多工具代理。
development
airunway-aks-setup
microsoft
在AKS上設定AI Runway——從裸叢集到執行模型。涵蓋叢集驗證、控制器安裝、GPU評估、供應商設定及首次部署。時機:「設定AI Runway」、「上線AKS叢集」、「安裝AI Runway」、「airunway設定」、「部署模型至AKS」、「在AKS上進行GPU推論」、「在AKS上設定KAITO」、「在AKS上執行LLM」、「在AKS上使用vLLM」、「在AKS上設定模型服務」、「AI Runway控制器」。
devops
appinsights-instrumentation
microsoft
使用Azure Application Insights檢測Web應用程式的指南。提供遙測模式、SDK設定與組態參考。適用時機:如何檢測應用程式、App Insights SDK、遙測模式、什麼是App Insights、Application Insights指南、檢測範例、APM最佳實踐。
devops
applicationinsights-web-ts
microsoft
使用Application Insights JavaScript SDK(@microsoft/applicationinsights-web)為瀏覽器/Web應用程式進行檢測。適用於真實使用者監控(RUM)——頁面檢視、點擊、AJAX/fetch依賴、例外、自訂事件,以及與後端OpenTelemetry追蹤關聯的瀏覽器端GenAI代理追蹤。涵蓋SDK載入器指令碼與npm設定、框架擴充(React、React Native、Angular)、點擊分析、遙測初始化器,以及從瀏覽器發出的代理/工具/模型span的OTel GenAI語意慣例。
devops
azure-ai-anomalydetector-java
microsoft
使用適用於 Java 的 Azure AI 異常偵測器 SDK 建置異常偵測應用程式。在實作單變量/多變量異常偵測、時間序列分析或 AI 驅動監控時使用。
development
azure-ai-language-conversations-py
microsoft
使用 azure-ai-language-conversations Python SDK 實作對話語言理解(CLU)。當使用 ConversationAnalysisClient 分析對話意圖與實體、建置 NLP 功能,或將語言理解整合至應用程式時使用。
development
azure-ai-ml-py
microsoft
Azure Machine Learning SDK v2 for Python。用於機器學習工作區、作業、模型、資料集、計算資源與管線。 觸發詞:「azure-ai-ml」、「MLClient」、「workspace」、「model registry」、「training jobs」、「datasets」。
development