The sections have been modified a fair amount. The Entries menu has been extended to have: Fixing entries, Updating author info, Thanks for the help. The first two were moved from FAQ and the last one from About. The rationale for these, in order: entries and authors have to do with entries and most of the help involves entries. It also has to do with website but in 'about' it does not seem the right place, unless maybe it is renamed to 'Credits' or something (I am not suggesting this one way or the other). This one is debatable but that was the rationale. The Status menu has been extended to have: Markdown guidelines and Security policy. The rationale is that the markdown guidelines are part of contest guidelines and the security policy is about the status of the entries and how if there are any security problems they will not be addressed (instead of about the IOCCC). The FAQ was extended to have: Compiling entries, Running entries, How to help. The reason to not have compiling and running entries under Entries is because this is 'how to' and also the menu was too long. How to help seems like a good idea under FAQ too (was in About). About was extended to have the new 'about.html' page (from about.md). This is a rare case where a new file is useful as it fits under 'About'. The menu now is: Home page, About the IOCCC, The Judges, Contact us. I was pondering the 'About [the IOCCC]' at the top and also naming it 'About us' but it seemed like a link to index.html was necessary and as for 'about us' it seemed less clear as to whether it was about the judges or the contest itself. The FAQ has been updated slightly for this. After further inspection the bin/index.html and the inc/index.html links were removed from the about section. With these changes, renaming an item or slight reordering notwithstanding, I believe that #2006 is probably in good shape.
To build:
make
There is an alternate version for those who wish to use QEMU rather than having to reboot and rely on having a floppy drive (if you even remember what those are :-) ) etc. See Alternate code below.
Bugs and (Mis)features:
The current status of this entry is:
STATUS: INABIAF - please DO NOT fix
For more detailed information see 2004/gavin in bugs.html.
To use:
./gavin
Try:
To enjoy the results (on a Linux/x86 machine) with a floppy drive (remember those? We do!):
su
Make a floppy with an ext2 filesystem (mke2fs /dev/fd0 replacing fd0 with
whatever the floppy device is!), mount it under /mnt/floppy and then:
cp kernel fs.tar lilo.conf boot.b /mnt/floppy
NOTE: If the version of your lilo is not 21.4, use the appropriate boot.b file.
NOTE: If your floppy drive is not /dev/fd0, edit lilo.conf appropriately.
lilo -C /mnt/floppy/lilo.conf
Boot from the floppy on an x86 machine with a PS/2 keyboard and mouse. Move the window away from the corner.
vi gavin.c
Scroll up and down the file with up-arrow and down-arrow.
sh
You can have several shell windows:
vi README.md
./prim
Press and hold any key.
Alternate code:
This version's main difference is that the macro K is redefined to 1 so that
you can use it with QEMU.
Alternate build:
make alt
Alternate use:
The use is mostly the same as gavin except that one initially executes gavin.alt
and one will have to use QEMU instead. The files
generated are the same names. See to use and try above as
well as the judges' remarks below plus gavin.html.
Judges' remarks:
Over the years, we've seen a program that immediately dumps core if executed on the "wrong" architectures (e.g., 1984/mullender entry) and a program that dumps core after a while (e.g., 2001/bellard entry). This entry will happily compile and execute on most any architecture - the trick is that the result, to be useful, must be produced (and can only be enjoyed) on a particular architecture. We have not decided yet whether to disallow this for future contests.
If you do not want to mess with a floppy and you use GRUB, see gavin.html.
You can put additional text files in fs.tar for browsing with vi.
If you do not want to bother rebooting your computer at all, see https://www.qemu.org for QEMU (by Fabrice Bellard, an IOCCC winning author) and use the Alternate code instead. You'll have to move the mouse to trigger the initial screen update.
The judges were able to write a few more programs to run in this OS. What are the limitations for such programs?
What can you do and what can you not do in such programs?
The program called prim, written by the judges, computes and prints prime
numbers.
Author's remarks:
This is a 32-bit multitasking operating system for x86 computers, with GUI and filesystem, support for loading and executing user applications in ELF binary format, with PS/2 mouse and keyboard drivers, and VESA graphics. And a command shell. And an application - a simple text-file viewer.
The OS has ended up as a traditional monolith with one entry point for syscalls, with apps treated as cooperative-multitasking tasks that can be sent different messages (e.g. 'initialize', 'key hit', 'render to buffer') through their main entry point, and which will return control to the OS after having performed the necessary work. Applications are passed a pointer to their task structure, which is partially defined by the OS, and partially for the application's own use.
The program compiles into a tool to build a kernel image,
so having built the program, the Makefile will run it,
redirecting the output into a file called kernel.
The Makefile will then proceed to build a root filesystem image.
This involves rebuilding the program with different compiler flags,
then building a tarball containing the resulting programs
(the filesystem format supported by the OS is the tarball format).
For further usage information see gavin.html.
The filenames vi and sh are significant, and should not be changed.
Known 'features':
Known issues are really too plentiful to list.
If the mouse pointer goes off the left hand side of the screen it will reappear on the right, and vice-versa. If it goes off the top or bottom, it will go and corrupt some memory.
The file system is kinda optimistic about matching names, so,
for example if you type the command shell into a command-line
it will execute the program sh - close enough a match for it.
The ELF binaries are not loaded at the correct address, and their entry point must be the start address of the text segment.
The keyboard driver can cope with the basic alpha-numeric keys,
but gets confused by fancy things like shift or backspace.
In the text-file viewer, vi, the up/down and PgUp/PgDn keys
scroll up or down by one line. There is nothing to stop you
from scrolling above the top of the file, and pressing any
other keys might have an undefined effect.
The x86 is bootstrapped into 32bit mode in 6 instructions, with 4 more to set up
data/stack segments and a stack-pointer to allow C code to be run. On top of
this there are also about a dozen instructions to switch the video card into
graphics mode. All in all, a relatively tiny number of instructions next to the
size of the C program. Also, the string is mostly composed of data - a
Linux-esque kernel header for the bootloader, protected mode descriptor tables,
keyboard maps, etc. (I should also mention that it contains mini functions to
perform an x86 in and out instruction - to allow the keyboard & mouse to be
driven from C code).
Porting to another architecture should be relatively easy1 -
the string simply needs be replaced with one containing
data & code suitable for the new target platform.
Accesses to data in the string are made relative to the define V,
so these may need updating as appropriate (0x90200 is the address
at which a Linux bootloader loads an x86 kernel image).
-
* ;-) ↩︎