Submit: Turn in your
kprintf.c and testcases.c
source files using the
turnin command on morbius.mscsnet.mu.edu or
one of the other
Systems Lab machines.
Work should be completed in pairs. Be certain to include both names in the comment block at the top of all source code files. It would be courteous to confirm with your partner when submitting the assignment. You may modify any files in the operating system, but only changes to "kprintf.c" will be graded for this assignment.
You will have to familiarize yourself with several common UNIX tools for this assignment. The first of these is tar, a utility originally devised to create tape archives for the purpose of backing files up onto computer tapes.
While tar is still used to create tape backups of file systems, it has become far more common to use tar to group files and/or directories together into a single entity, typically called a "tar-ball." (So common is the use of tar that it has been verbed in computer science terminology: We speak of "tarring" files, or files that have been "tarred up.") Tar syntax is somewhat arcane, as tar came into existence before modern standards for command-line options.
This untars the files into your working directory, in a subdirectory called xinu-hw3.
For more information on tar, please see the UNIX man pages.
While the gcc command-line options provide a great deal of flexibility when compiling programs, things quickly become unmanageable when the number of source files exceeds what you can conveniently type in a few seconds.
The make utility can be thought of as a companion to the compiler infrastructure (preprocessor, compiler, assembler, and linker) that allows the build rules for large projects to be explicitly encoded in Makefiles. A Makefile typically consists of common definitions, (such as, which compiler to use), and a set of rules. Each rule has a target, such as the file that is to be built, and can be followed by a list of dependencies and a sequence of steps to perform in order to build that target. In addition, make has quite a few common rules built into it.
You will not have to write your own Makefiles for this course, but you will have to use and possibly modify some for all of our remaining assignments. The Makefile is always human-readable, so feel free to open them up and look around.
To build the XINU operating system, perform the following steps:
This should produce about a page of output as each source file is compiled, and the resulting object files are linked together to form the operating system, a simple set of library functions, and the boot loader. If all goes as it should, you should find the directory full of .o files from all of the source code in the other subdirectories, and most importantly, a newly compiled operating system image called "xinu.boot."
For more information on make, please see the UNIX man pages.
Your XINU image is now ready to be run on a backend machine. To
transfer it there, we have a special utility
called riscv-console. Execute riscv-console in
the compile directory where your
xinu.boot file resides. riscv-console will
connect your terminal to the first available backend machine, and you
should see a message like:
connection 'poodoo', class 'riscv', host 'morbius.mscs.mu.edu'
depending on which backend you get. This will be immediately
followed by a stream of automated commands as the embedded target
system boots, configures its network settings, and uploads
your xinu.boot kernel.
The most important thing to remember about riscv-console is that it is modal, like vim. You start out in direct connection mode, in which your terminal connects directly through special hardware to the serial console on your backend machine. To get out of riscv-console, hit Control-Space, followed by the 'q' key.
The source tar-ball we are starting with contains only a few files for the operating system proper, in the subdirectory system. We will be adding files into this directory in every subsequent assignment.
The other files in the XINU subdirectories break down as follows:
Your task for this assignment is to write a simple synchronous serial driver for the embedded operating system, so that you can see what you are doing in all subsequent assignments.
The driver is "synchronous" because it waits for the slow I/O device to do its work, rather than using interrupts to communicate with the hardware.
The driver is "serial" because it sends characters one at a time down an RS-232 serial port interface, like the one found on most modern PC's.
The driver is a "driver" because it provides the software interface necessary for the operating system to communicate with the hardware which, in this case, is an I/O device.
This platform's serial port, or UART (Universal Asynchronous Receiver / Transmitter) is a member of the venerable 16550 family of UARTs, documented here. Of particular interest to us is section 9.2.5 of the specification, which describes the registers accessible to programmers. On this platform (the Sipeed Nezha), the UART control and status registers are memory-mapped, starting with base address 0x2500000. You can view these address definitions in include/ns16550.h
The file system/kprintf.c has the skeleton code for four I/O-related functions: kputc(), (puts a single character to the serial port,) kgetc(), (gets a single character from the serial port,) kungetc(), (puts "back" a single character, ala K&R's getch() and ungetch() functions,) and kcheckc() (checks whether a character is available.) Each function contains a "TODO" comment where you should add code. The actual kprintf() is already complete; it will begin working as soon as you complete the kputc() function upon which it relies.
[Revised 2023 Jan 31 02:35 DWB]