Скачать или смотреть Solving a Shared Memory Issue in Multi-Process C/C++ Programs

Debugging multi-process C/C++ applications is hard — especially when tackling race conditions or tricky timing issues involving shared memory.

In this video, Isa from Undo walks you step-by-step through how to reproduce, record, and diagnose a real shared memory race condition using Undo's technology: LiveRecorder and UDB.

🔍 What you’ll learn:

How to record multi-process programs using --on-fork both

How to track shared memory (shmem) access with --mpc-shmem

How to replay failures deterministically in UDB

How to navigate processes with multi-process correlation

How to use ublame to inspect shared memory read/write history

How to identify and fix an unsafe read-modify-write pattern

How to debug faster without adding printfs or rebuilding

🧠 Who is this for?

Developers working with:
• C/C++ multi-process systems
• fork(), multiprocessing, IPC
• shared memory regions
• timing-sensitive or non-deterministic bugs
• debugging race conditions

🚀 Tools featured:

• LiveRecorder (Undo’s recording engine)

• UDB (Undo’s reversible debugger)

• multi-process correlation (MPC) for shared memory: https://docs.undo.io/MPCSharedMemory....

• on-fork: https://docs.undo.io/MPCSharedMemory....

With Undo, you can record all processes at once and instantly trace the true root cause of any bug.

If you’ve ever struggled debugging multi-process systems in C/C++, this video is for you!

👍 Like the video if it helps
💬 Comment your debugging questions or suggestions
🔔 Subscribe for more deep-dive debugging tutorials