My CppCon 2019 talk, If You Can’t Open It, You Don’t Own It is now up on YouTube.

Anthony Williams, author of “C++ Concurrency in Action: Practical Multithreading“, was in the audience and mentioned the talk in CppCon 2019 Trip Report.

Abstract:

For the past 30 years, we have dealt with penetrations into secure systems almost exclusively from the software layer: applications and operating systems. With the advent of side channel exploits like Spectre, Meltdown and Foreshadow, hardware designs are now battlefields. In this talk, we’ll look at four real-world hardware attacks that changed the way we think about secure systems and see how hardware exploit strategies drive software exploit strategies.

And what that means for the future of Modern C++.

We’ll explore four lines of attack:

• Roots of Trust,
• Side channels exploits,
• How physical access creates opportunities, and
• How our supply chains often create our greatest vulnerabilities.

As the Standards Committee puts the final touches on C++20 this year, we’ll use these as the framework to get an inside look at the committee’s efforts to build a safer, more resilient language. We’ll see:

• How new language features, like Concepts, Contracts and Ranges, help (or hurt) our ability to write secure software.
• How Undefined Behavior is explicitly used by compiler developers to generate high performance machine code and what that means for software security.
• Which proposals coming for C++23, like Zero-overhead deterministic exceptions and secure_clear, will help address some of the worst vulnerabilities in the language.

This talk is about how our language and design choices affect our system’s ability to withstand attack. It’s also about how the evolution of the language is addressing the insecure world it operates in and the places where it still falls short.

Eratta:

Near the end of the talk I was discussing Concepts and made the comment about alternatives using asserts being a run-time check for the correctness of the types. My slides have static_asserts which are clearly compile time checks. This is what happens when you make a last minute slide change to “improve” the point but you can’t see the slide with the changes you made.

Sorry for the confusion.