Skip to main content

Checkpoint Gate

World-Class Evidence Integration

The checkpoint must verify that the World-Class Strand evidence exists, is reproducible, and has a written highest-risk gap.

Required Output Classification​

Required outputClassificationPublic/private guidance
Closed-book prompts, self-assessment answers, and skills matricesPractice artifactUse for honest calibration; do not publish raw answers unless rewritten as a study guide.
Required evidence gate items, sign-off checklist, and readiness decisionCheckpoint evidenceKeep as private progression evidence; share only sanitized summaries with mentors or reviewers.
Repair artifacts produced after a weak checkpoint, such as corrected solutions, diagrams, traces, benchmarks, or runbooksCheckpoint evidenceStore beside the checkpoint so the remediation trail is inspectable without making mistakes public.
Reviewer notes or mentor feedback that materially improve a project artifactPortfolio candidateConvert into public-safe acknowledgements or changelog entries only after removing private feedback context.

Use this gate before moving to Semester 6. The standard is not "I saw the material once." The standard is "I can reason through it from memory and support claims with traces, measurements, or working code."

How To Use This Gate​

For each item, mark one of:

  • Confident if you can answer or perform it now, from memory
  • Review needed if you would need notes, prompts, or another practice pass

If two or more items in the same module are weak, return to that module before advancing.

Knowledge Checks​

Module 1: Processes & Scheduling​

  1. I can compare FCFS, SJF, SRTF, round-robin, and MLFQ on a small workload and compute turnaround and response time. [ ] Confident [ ] Review needed
  2. I can explain a context switch step by step, including why cache and TLB effects matter. [ ] Confident [ ] Review needed

Module 2: Memory Management & Virtual Memory​

  1. I can walk a virtual address through a multi-level page table and distinguish TLB miss from page fault. [ ] Confident [ ] Review needed
  2. I can explain replacement-policy tradeoffs, allocator roles, and when mmap or CoW changes performance behavior. [ ] Confident [ ] Review needed

Module 3: Concurrency & Synchronization​

  1. I can write the interleaving that breaks naive shared-memory code and explain the smallest correct fix. [ ] Confident [ ] Review needed
  2. I can use mutexes, condition variables, semaphores, and lock-ordering rules without confusing safety, liveness, and fairness. [ ] Confident [ ] Review needed

Module 4: File Systems & I/O​

  1. I can explain what write() returning does and does not guarantee, and how journaling or CoW changes crash outcomes. [ ] Confident [ ] Review needed
  2. I can justify when to choose blocking I/O, epoll, or io_uring for a service. [ ] Confident [ ] Review needed

Module 5: Network Protocols & Sockets​

  1. I can do subnet reasoning, explain TCP state transitions, and distinguish flow control from congestion control. [ ] Confident [ ] Review needed
  2. I can write and debug a socket server using packet and socket-state tools rather than guesswork. [ ] Confident [ ] Review needed

Skills Checks​

  1. I have at least one working scheduler, memory, concurrency, filesystem/I/O, or networking artifact that I can run and explain. [ ] Confident [ ] Review needed
  2. I have used real inspection tools this semester such as perf, strace, /proc, ss, tcpdump, or Wireshark and wrote down what I observed. [ ] Confident [ ] Review needed
  3. I can take one bug or performance complaint and turn it into a concrete hypothesis, measurement plan, and written conclusion. [ ] Confident [ ] Review needed
  4. I can explain how this semester feeds directly into databases and distributed systems next semester. [ ] Confident [ ] Review needed

Required Evidence Before Advancing​

You should be able to point to:

  • one process or scheduler artifact
  • one memory or page-fault measurement note
  • one concurrency exercise with a documented broken interleaving and fix
  • one filesystem or I/O measurement or crash-consistency note
  • one network capture or working socket service

If you do not have evidence for one of these areas, the semester is incomplete even if your notes look good.

Common Blockers To Resolve First​

Do not move on if any of these are still true:

  • you confuse a TLB miss with a page fault
  • you cannot produce a deadlock example and name the condition your fix removes
  • you treat write() success as "durable on disk"
  • you cannot explain TIME_WAIT, CLOSE_WAIT, or why epoll scales better than select
  • you have no packet capture, no perf output, and no actual measurements from the semester

Sign-Off​

  • I completed the cumulative review
  • I completed the semester exam
  • My semester project and module artifacts are runnable or clearly documented
  • I am ready to begin Semester 6: Databases & Distributed Systems

Mastery Rubric​

LevelEvidence
Beginner passCan answer direct questions and complete familiar exercises with light notes.
Solid passCan solve new variants, explain choices, and connect the work to Semester 4 Systems Programming.
Strong passCan defend tradeoffs, identify failure modes, and produce clean evidence in the portfolio artifact.
Not readyRelies on copied solutions, cannot explain mistakes, or lacks durable artifacts.

Retake and Repair Rule​

If a section is weak, do not only reread. Repair it by producing new evidence: a corrected solution, a fresh implementation, a rewritten proof, a benchmark, a diagram, a runbook, or a short teaching note.

Answer-Quality Examples​

Use these examples when grading written answers or spoken explanations.

QualityExample pattern
WeakNames a concept but gives no example, constraint, or failure case.
AcceptableDefines the concept and applies it to a familiar exercise.
StrongApplies the concept to a new variant and explains why an alternative would fail.
Portfolio-readyConnects the concept to Semester 4 Systems Programming, current project evidence, and a future capstone decision.

Interleaving Prompt​

For any missed answer, add one sentence starting with: This depends on an earlier skill because...​

Calibration Materials​

Use these learner-visible calibration materials before self-grading or requesting review:

Scheduling-contract evidence​

Require every PS1–PS6 gate at level 2 or above. Reuse the simulator packet: event-order predictions, independent metrics, queue-policy fault evidence and a bounded quantum/cost comparison. A simulator result does not establish host scheduler behavior. Failed gates remain incomplete regardless of aggregate score. See the reconciled workload and progression contract.

Translation-contract evidence​

Require every VM1–VM6 gate at level 2 or above. Reuse address traces, permission/fault cases, ASID separation, remap/invalidation tests and counter reconciliation. A TLB hit does not guarantee authorized access, and a miss does not imply storage I/O. Failed gates remain incomplete regardless of aggregate score.

Queue-concurrency evidence​

Require every CQ1–CQ6 gate at level 2 or above. Reuse protected-state reasoning, controlled wait/close traces, fault/repair evidence and the independent shutdown case. Sequential model histories and sampled thread tests do not prove every interleaving or fairness. Failed gates remain incomplete regardless of aggregate score.

Crash-recovery evidence​

Require every FS1–FS6 gate at level 2 or above. Reuse the crash matrix, persisted-prefix cases, early-success counterexample and repeatable replay evidence. Model results do not certify real device durability. Failed gates remain incomplete regardless of aggregate score.

Stream-framing evidence​

Require every NW1–NW6 gate at level 2 or above. Reuse fragmentation/EOF traces, early size rejection, terminal-error policy, simulated deadlines and the bounded socket-adapter design. Decoder tests do not establish real network timing or exactly-once business effects. Failed gates remain incomplete regardless of aggregate score.

Semester completion contract​

Use the 468–676-hour workload and progression contract and the five separately scored security-strand categories S1–S5. All 34 module gates (including PC1–PC4) and all five strand categories must reach level 2; aggregate marks cannot replace missing evidence. Reuse artifacts and record time once. Independent review and workload pilots remain pending.

Native parallel-work evidence​

Require PC1–PC4 at level 2 alongside existing gates. Reuse the native histogram packet during semester review; do not add parallel execution to the service without a workload justification. Clinic time is included in the revised 468–676-hour semester estimate.