Introduction to Systems Programming

Systems programs form the infrastructure of computing: Operating Systems, device drivers, Network Protocol Stacks
Key characteristics:
- Strict performance constraints
- Need for Memory Management efficiency
- Require control over data representation
- Need to effectively manage O Operations
- Must handle Shared State effectively

Systems programming requires low-level control over data layout and memory
Systems components are often bottlenecks as they form the foundation for higher-level components
Languages like C/C++ remain popular for systems programming due to the control they provide
Languages such as Java, Python, and Go lack the necessary low-level control for true systems programming

Most modern systems run some variant of Unix (Linux, macOS, iOS)
Windows is an exception, being influenced by VMS
Unix and C have proven remarkably resilient, with core APIs recognizably the same after decades
Success factors include:
- Portability
- Source code availability to universities
- Relatively clean and consistent API
- The pointer abstraction in C provided a simple yet powerful mechanism

Unix Networking and Filesystem APIs can be performance bottlenecks
Security Models designed for multi-user systems don’t fit personal data protection needs
Lack of portable APIs for modern concerns (mobility, Power Management)
C is flexible but error-prone, making Security Vulnerabilities common
Limited support for Concurrency

End of Moore’s Law and Dennard Scaling
- Clock frequencies have stalled since mid-2000s due to power constraints
- Number of Cores increasing exponentially instead
- Parallelism becoming essential
Increasing Concurrency
- Multi-core Systems are now standard
- Concurrency Bugs becoming more visible
- Writing correct concurrent software is difficult
Security Concerns
- Always-on Internet Connectivity exposes vulnerabilities
- Number of security vulnerabilities increasing annually
- Memory Safety issues are a primary concern
Connectivity and Mobility
- Devices are mobile, battery-powered, and network-connected
- Cloud Computing and offloading computation is common
- Applications need to adapt to network variations

Need for safer systems programming languages that maintain control over data representation
Modern Type Systems can help improve expressivity, correctness, and security
Apply Functional Programming techniques to systems programming
Use Type-Driven Development to catch issues at compile time

Quartz 4