3 * The tests in tests/ don't apply the grading patches.
5 * We need better example programs.
9 - join-invalid doesn't compile if tid_t is not scalar type.
11 - mlfqs tests suck. They aren't even correct, e.g. the amarv
12 submission from win0405 is graded incorrectly.
16 - Don't emphasize that stuff needs to be copied from user space to
17 kernel space. Instead, emphasize validation and suggest that
18 copying is a common solution and that it will be necessary in
19 project 3 and in real OSes. Also revise the grading criteria to
22 - Move `join' implementation here, from `threads' project, to help
23 normalize the project difficulties.
25 - The semantics of the join system call should change so that it
26 only returns the exit code once.
28 - Mark read-only pages as actually read-only in the page table. Or,
29 since this was consistently rated as the easiest project by the
30 students, require them to do it.
32 - Don't provide per-process pagedir implementation but only
33 single-process implementation and require students to implement
34 the separation? This project was rated as the easiest after all.
35 Alternately we could just remove the synchronization on pid
36 selection and check that students fix it.
40 - Increase maximum disk size from 8 MB to something that actually
41 requires doubly indirect nodes. There is a negative pressure here
42 from the bitmap object--perhaps we need a specialized bitmap that
43 doesn't have to be all in-memory at once.
45 Alternatively, shrink the inode size.
47 - Add mkdir and ls example user programs.
49 - Add option to disable buffer cache.
51 - Get rid of "dump" commands--they're not really useful.
55 - Finish writing tour.
57 - Add "Digging Deeper" sections that describe the nitty-gritty x86
58 details for the benefit of those interested.
60 - Add explanations of what "real" OSes do to give students some
67 . Low-level x86 stuff, like paged page tables.
71 - mmap/munmap should use segment IDs like Nachos. Too hard
74 - Add src/testcases/vm, src/testcases/filesys and make it clear to use
79 - Release some of them.
81 - The threads, userprog, vm test source files could use
82 factorization and cleanup along the lines of fslib in the filesys
85 - The p1-4.c testcase needs significant tuning. Currently it takes
86 too long (especially when SHOW_PROGRESS is turned on) and doesn't
87 show significant improvement.