C++ Study Questions#

• What are some common examples of undefined behavior?
  • What is memory corruption?
    • What are some common causes of it?
  • How can you get UB without memory corruption?
  • How does UB interact with optimization?
• What is RAII? How does it differ from garbage collection?
• Why use smart pointers instead of raw pointers?
• What is the STL?
  • What is it useful for?
  • What are some common STL collections?
  • What are some common STL algorithms?
  • How do iterators work?
  • Why can’t you dereference the end iterator?
• What is the difference between “value semantics” and “reference semantics”?
• What is “type erasure”?
  • How is std::function implemented?
  • How is std::any implemented?
    • What does std::any even do?
• Why does an empty struct have a size of 1 and not 0?
• What is std::atomic?
  • How does it differ from volatile?
  • How can it be optimized?
• What are the versions of the C++ standard?
  • What major features were added in each of them?
• What are the major C++ compilers?
  • Under what licensing terms are they available?

C++ Performance#

• What are some common compiler optimizations?
  • What are some optimizations a compiler cannot do?
  • What are downsides to having the compiler do optimizations?
• Why are virtual function calls slower?
• What parts of a codebase need most optimizing?
  • What parts do not?
• Common compiler flags
  • What is the difference between -O2, -O3 and -Os?
  • What is -g?
    • Why is it normally combined with -O0?
      • What does -O0 do?
      • How is that different from what -g does?
• What is the difference between throughput and latency?

Operating System Questions#

• What is the difference between the stack and the heap?
• What is an operating system kernel?
  • What are some examples of operating system kernels?
  • What is “kernel” or “supervisor” mode?
  • What is a monolithic kernel vs a microkernel?
  • What is a driver?
    • Can they run in usermode?
      • When?
• What is a system call?
  • What is the difference between a blocking and non-blocking system call?
• What is the difference between a process and a thread?
  • What are the performance implications?
• Describe different forms of IPC
  • What is the difference between brokered and non-brokered IPC?
  • What considerations should you take into account when choosing an IPC system?
• Describe virtual memory.
  • What is an address space?
  • What is memory protection?
  • What is paging?
    • What is a page fault?
  • What does a kernel have to do to implement virtual memory?
  • How does a program allocate more memory on the stack?
  • How does a program allocate more memory on the heap?
    • In userspace?
    • What syscalls might it have to make?
  • How do memory mapped files work?
    • When should you prefer this to read and write syscalls?
  • How does shared memory work?
    • When should you prefer this to other forms of IPC?
  • What is swap?
  • What is a TLB?
    • What performance implications does it have?

Network Study Questions#

• Explain the layers of the OSI model
  • How do they map to Internet-based protocols?
• Explain IP basics
  • What is IP?
  • What is an IP address?
    • What is a subnet?
    • What is a non-routable IP address?
    • What is localhost?
      • When would we use it?
    • What is NAT?
      • When is it used?
  • How does IP relate to link-level protocols?
    • Give some examples of link-level protocols.
    • Explain CSMA/CD.
    • What is MTU?
    • Explain Path MTU Discovery, and why it’s important.
  • What is ICMP?
    • What are 2 command-line utilities that use it?
    • Why is it important?
  • What is the difference between a hub, a switch, and a router?
  • What is the difference between IPv4 and IPv6?
• Explain DNS basics
  • What is DNS?
  • What are some types of DNS records?
  • On Unix
    • What is /etc/hosts?
    • What is /etc/resolv.conf?
    • What sort of things can go wrong if these are configured incorrectly?
  • How do you access DNS from your applications?
• What is a VPN?
• Explain the basics of TCP.
  • How does TCP implement reliability on top of IP?
  • TCP is stream-oriented, UDP is packet-oriented. What does this mean?
  • Explain the TCP three-way handshake.
    • What are SYN cookies?
  • How do TCP acknowledgements work?
    • How does TCP handle the lack of negative acknowledgements?
  • What is the TCP window size?
• Explain the basics of UDP.
  • What are some differences between TCP and UDP?
  • Why might you want to use UDP?
  • How are UDP broadcasts implemented?
• On Linux, what sort of things can you tune about networking using sysctl?
  • Why might you want to do this?
• Why are heartbeats important?
  • Why might they be implemented in user protocols?
    • Why is TCP keepalive not enough?

Network Programming Practice Projects#

• Echo server
  • Can test by hand using netcat
  • v1: Accepts single TCP connection, echos all inputs
  • v2: Use threading to accept multiple connections, echo each to itself
  • v3: Use single-thread
    • select or epoll syscall
    • async in Rust
    • Network reactor/dispatcher library in C++
• Chat server
  • Accept multiple connections
  • Can test by hand using netcat
  • v1: Send all whole lines (requires buffering) to all clients
    • No length restriction
  • v2: Handshake to set username
    • Make still it still is testable using netcat
    • Server sends username along with messages
• File server
  • Simple HTTP-like protocol to specify filename of what file to send
    • Make sure you sanitize for inputs!
  • v1: Serve one static file at a time
  • v2: Stream multiple files to multiple clients without using threads
  • v3: Support uploads
    • Write custom client
• Exchange
  • v1: Just exchange connection
    • Keep order books
      • When new order comes in compatible with old order, make trade
      • One security at first, then multiple
        • But build multiple into protocol right away
    • TCP Protocol:
      • Create a custom protocol with CLI client
      • Client -> Server
        • Orders
          • Only GTC limit orders at first
          • Quantity
          • Security
          • Price
          • Buy/Sell
          • Order ID
            • Why is this useful?
      • Server -> Client
        • Order confirmation
          • All order information, echoed
        • Trade confirmation (when orders match)
          • Also indicate whether client was maker or taker
  • Add-ons
    • Add heartbeats to protocol
    • Market data
      • UDP broadcast
      • Current state of order books
    • Drop copy
      • Separate TCP records of all trades
      • Record trades in file before confirming
      • Allow replay of previous days' trades
    • Fees and kickbacks
      • Charge takers
      • Reimburse makers
      • Configurable on by-security basis
    • Credit limits
    • Advanced order types
      • IOC/FOK
      • Maker-only orders
    • FX-style last look
    • SSL support