Skip to content

bug_log.md

Latest commit

 

History

History
255 lines (185 loc) · 7.85 KB

bug_log.md

File metadata and controls

255 lines (185 loc) · 7.85 KB

BUG LOG

2014-05-05

  • Fix permissions for video memory

    • Prior to this any user-level program could write to video memory without repercussion, arguably not notable, but an issue nonetheless
    • Fixed by refactoring functions to allow flags

  • Fix issues related to halting the last process in the OS

    • Exiting from the last shell would cause a loop of pagefaults from user memory
    • Fixed by spawning a new shell after the last process closes

  • Fix index out of bounds that caused memory issues related to fish on multiple terminals

    • Fish would appear on other terminals and in general wreak havock if more than one ran at once
    • Fixed by fixing an array index that often appeared out of bounds

  • Fix issues related to killing processes with crtl-c

    • Processes would sometimes get removed from the scheduling queue when they shouldn't
    • Fixed by refactoring a number of commands to directly accept pointers to pcb

  • Fix sys_halt_internal implementation and issues with RTCs closing when they shouldn't

    • A bug in rtc_close often removed the wrong rtc from the global rtc chain, or otherwised caused other process's rtcs to stop updating

  • Fix up some code related to sys_halt

    • Refactored sys_halt to be useable inside the kernel, which fixed a wrong return code issue when a program was squashed after generating an exception

2014-05-04

  • Fix some rtc bugs, halt when last shell is closed

  • Fix page fault on switching terminals
    • Page fault would occur when fake memory became unmapped. Characters would try to write to nonexistent video memory.
  • Fix issue where switching terminals didn't actually switch the correct terminal
    • The values for the terminal were based on the current process on the scheduler. We had to use the correct value in the switch terminal.
  • Fix issue where we didn't switch to a page directory containing fake video memory before attempting to restore the screen
    • The video memory could not be drawn to the screen, because the pointers were all wrong.

2014-05-03

  • Fix issue where terminal cursor didn't wrap to the previous line when backspacing
    • There was an issue with where we updated our cursor.

2014-05-02

  • Fix forward declarations

  • Fix issue with mapping user video memory
    • The pointers to video memory were all 64kB off, because a 1 was added for no obvious reason.
  • Fix scheduling by removing actively running process from queues.
    • The scheduler was not removing running processes correctly.

2014-05-01

  • Fix issue where schedule() is only called once
    • The PIT was not calling the scheduler after the first call, so we had to make sure the variables were not null.
  • Fix use of uninitialized variable in scheduler
    • We initialized our variables before calling them.
  • Minor fixes
    • Miscellaneous bugs and spacing/new lines were formatted.
  • Fix permissions
    • We wanted the user to only receive user permissions, but he was able to do kernel things.
  • Fix last issue where screen cursor doesn't sync to global cursor
    • The screen_x and screen_y variable was not accounted for.
  • Fix issues related to improperly assigning video memory
    • Pointers were pointing to invalid memory locations, causing page faults.

2014-04-27

  • Fix #11, where backspacing after clearing the screen caused a page fault."
    • The terminal would try to write before video memory.
  • Fix header issue where static inlines needed global variables
    • Had to reformat the location of static inlines.
  • Fix issues relating to running the OS on real hardware.
    • We made a bootable USB, but it did not work at first. We had to make the OS wait for the keyboard.

2014-04-22

  • Fix issue with virtual rtc where using frequency macros resulted in incorrect frequencies
    • The macros were defined incorrectly, causing strange frequencies as a result.
  • Fix bug where virtual rtc would tick immediately upon being opened
    • The RTC was ticking too soon, so it needed an initial condition to prevent it from happening.

2014-04-21

  • Fix issue with clearing/setting interrupts
    • Interrupts were not being set correctly, and clearing would break the OS.

2014-04-20

  • Fix issues that arose from custom filesys_img, copy script from class directory that does it for us properly.
    • We modified file_sys and strange things happened, so we had to modify our code slightly.

2014-04-19

  • Fix keyboard initialization on real hardware
    • Made a while loop to ensure the keyboard was ready.
  • Remove unused code and fix character
    • Some outdated code was floating around in our directory.
  • Fix keyboard buffer handling when we're not in a term_read
    • Keyboard buffers would max out if they werent read.

2014-04-18

  • Fix bugs related to spawning new proceeses and address calculation
    • Fixed pointer issues for spawning new processes.
  • Fix bug in term.c where keys are echoed to the screen even when they're not added to the key queue
    • Characters were printing regardless of their existence in the terminals queue.

2014-04-17

  • Fix '{' bug in PIC
    • The PIC would display a bracket on the screen due to a syntax error.
  • Fix bug where userspace address was not properly checked; user programs now squashed on raised exception
    • The user space was not treated differently than the kernel space initially.
  • Fix term_read bug related to uninitialized variables
    • Multiple newlines would be read from the end of the buffer due to an uninitialized variable
    • Fixed by initializing the variable
  • Fix read_data, hopefully for the last time
  • Temporarily fix issues caused by reading blocksize of 2^n in read_data
  • Fix small style issues in syscall_impl.c and a bug where vidmapping didn't actually load pages for video memory
  • Fix paging bug where certain page tabes were RO when they should have been RW
  • Fix rtc driver implementation so that takes ptr to int instead of int
  • Fix issues related to compiling sample programs
  • Fix bug where syscall number bounds are not properly validated
  • Fix read_dentry_by_name for real this time
  • Fix issues with dir_read implementation

2014-04-16

  • Validate fd before calling read/write/close. Fixes #6.
  • Fix a bug where attempting to execute a non-executable would leave the system in a state where no further executables could run
  • Fix issue in read_dentry_by_name where we'd match filenames longer than the given string
  • Fix typo
  • Fix bug in read_data, where there were instances of offset that were not multiplied by db_first
  • Fix exec/halt implementation. Second process can now run more than once without crashing.

2014-04-14

  • Fix issues brought up by last commit
  • Fix read_data
  • Fix proc.h misc issues
  • Fix file_loader
  • Fix additional page directories

2014-04-13

  • Fix compile errors

2014-04-02

  • Fix linking issue with gcc by substituting ld
  • Fix dumb mistake where ports are not the same size
  • Fix portability issue related to using GCC for linking that caused the filesystem image to appear too high in memory

2014-04-01

  • Fix test index
  • Fix read test
  • Fix Filename errors
  • Fix implementation of term_write so that it writes the number of bytes passed

2014-03-23

  • Fix email script so that it rewrites by name as well.
  • Add script to fix emails in history. Dangerous. Do not use.

2014-03-20

  • Fix small spelling error
  • Fix authors.txt for my email
  • Fix incomplete variable name changes

2014-03-18

  • Fix RTC
  • Fix PIC init, reenable RTC, fix PIC EOI, implement IRQ handlers

2014-03-17

  • Fix spelling error
  • Fix exception handling; initialize IDT; define halt; halt on unhandled exceptions
  • Fix i8259_init

2014-03-16

  • Fix bug related to limit on GDT