Bug log

All the issues on Github will have details about all the bugs. Here we just listed those most buggy ones.

Checkpoint 1

No Bug, for real, cuz we're awesome :P

Checkpoint 2

1.Terminal Buffer #29

• Should ban user from typing after 128 words
• FIX: Disabled our Ring Buffer
• Nov 10th
• Fixed in 1 day

Checkpoint 3

1.Exception not always halt #37

• "Programs execute to completion, so you need not write a scheduler or deal with the timer chip yet, but you do need to be able to squash programs if they generate exceptions, returning control to the shell in such cases.", according to Page 8 in MP3 handout
• Nov 15th
• Fixed in 5 days

2.TSS.ESP0 should be maintained only by Scheduler #27

• "Everytime you do a context switch, TSS.ESP0 is NOT SAVED by Scheduler (a discrepancy to fix), thus sys halt still needs to update TSS.ESP0 THIS IS WRONG!! MUST BE UPDATED BY SCHEDULER!!! (IF nested interrupt happens during sys-halt: then the inner interrupt will already be using the stack we were trying to switch to!!!"
• Nov 16th
• Fixed in 1 day

Checkpoint 4

1.Getargs bug #42

• "If one types "cat frame0.txt", frame0.txt will be printed to screen, as expected. However, after that if someone types "cat" again, with no arguments cat will not exit with error and instead prints frame0.txt, which is not a good behavior."
• FIX: Deal with more corner case like this
• Nov 20th
• Fixed in 1 day

2.ls: verylargetxtwithverylongname.txt disappeared #41

• when ls, verylargetxtwithverylongname.txt disappeared
• FIX: For the last file we still need to get the file name, a simply if else corner case
• Nov 21st
• Fixed in 1 day

Checkpoint 5

1.Page fault after running many processes #48

• Happens after running 62 lss
• "Real reason: processDesc is messed up: (At every breakpoint, the first printed variable is theDispatcher, the second is processDesc)"
• Nov 28th
• Fixed in 2 days

2."ls" halt() crashes all the running processes, if long-lasting process exists. #63

• Reproducible by running either pingpong, or testvidmap.converted, or fish while repeatedly running "ls".
• Reason: when working on vidmap, we change the cpu paging but we forget to change back before sched
• FIX: change the cpu paging back to the process to be sched
• Dec 4th
• Fixed in 2 days

Extra

1.Moving mouse wrecks havoc in text mode #54

• Happens because text and video memory overlaps at the video card level
• FIX: Separate text mode/video mode mem explicitly
• Dec 3rd
• Fixed in 1hr

2.move mouse will be recognized as DOUBLE_CLICK #55

• Happens when we move the mouse, the mouse-handler need to be fixed
• FIX: Rewrite the logic for double click & single click
• Dec 2nd
• Fixed in 1 day

3.Map kiss fs to high virtual memory. #39

• "Calculate number of physical pages needed (~9 = 36mb) and map them to a continuous region in virtual memory. Gets rid of dirty paging hacks before reading files."
• FIX:"Gets rid of dirty paging hacks before reading files."
• Nov 16
• Fixed in 1 day

Bugs in TA's files

1.Cannot compile given code in lib.h

• located in lib.h
• cannot compile the code with outl()
• can compile before since it's lazy compilation, these part has never be used

Bugs in Makefile & Compiler

1.Makefile cannot find files in ./*/*

• Cannot find files in place such as subdirs/subdirs
• Fix with the help of StackOverFlow & Google, change Makefile largely
• Fixed in 7 days

2.NTFS cannot support symlink, which is required for our cross compiler

• When compile in network drive (NTFS on Windows), symlink is disabled
• The cross compiler require symlink, and wo it the compiler cannot work
• FIX: zip the compiler under *nix to tar.gz and extract it out, all the symlink will be extracted to things like dirs/subdirs
• Fixed in 7 days