On top of the computer hardware is operating system, utility programs, and application programs, which are the software component of a computer. The operating system manages the user/computer hardware interface, so that it is easier and friendlier to end users. The operating system supports the program creation, execution, manages the memory system, I/O devices and user processes.
UNIX operating system is a multi-user, multi-tasking operating system with an elegant hierarchical file system. The UNIX kernel does the real job of management. The user interacts with the kernel through a program called shell. Shell is just one of the many programs. There are several slightly different shells (sh: Bourne shell, csh: C shell, ksh: Korn shell, tcsh: Enhanced C shell).
A process in UNIX is a currently running program plus addition bookkeeping information. A process is created when you login. You can create process by running a program. Process A can create Processes B and C; Processes B and C is called children of A, A is parent of B and C. You can display the current running processes with the Unix command:
ps -aux (or ps -ef on other systems)or kill a process with the kill [pid] command.
Unix is a time sharing operating system. A typical workstation will have 30 to 40 processes running. But only one process is actually executing the instructions. The other processes are waiting for the CPU resources. The operating system controls which process should be executed and for how long. The operating system switches the running processes quickly. It gives user the impression that all processes are running simultaneously.
Unix has a large collection of commands (well over 300). It takes years of experience to be good at them. Here I list some of the interesting ones.
FILE SYSTEMS cd change directory cp copy file rm delete file mkdir/rmdir make/remove directory pwd print working directory ls list file df file system usage du disk usage quota or diskquota chmod change mode of the file protection ln create a link cat display a file diff print out the difference of two files EDITING, WORD PROCESSING vi tex/latex (document formating) emacs troff pico OTHER UTILITIES man display manual page grep pattern search awk pattern scanning and processing perl more advanced pattern scanning and processing dbx symbolic debugger make program maintenance spell spelling checker pine E-mail mail
floating-point
R-form Immediate Unsigned U/I single double C
add addi addu addiu add.s add.d "+"
sub subu sub.s sub.d "-"
mult/mul multu mul.s mul.d "*"
div divu div.s div.d "/"
Remarks: (1) Most arithmetic instructions take 3 operands, except mult
and div, which take 2. (2) Floating point instructions take only
3 floating-point registers. (3) First register is usually the result.
Result for mult and div is in hi and lo. (4) Immediate values take
the third operand. (5) What is the difference between signed and
unsigned version of the instructions? Note that the same "+" in C can be
translated into any one of the 6 different addition instructions.
Immediate C
and andi "&"
or ori "|"
xor xori "^"
sll "<<" shift left
srl ">>" shift right
sra ">>" shift right arithmetic
Remarks: (1) Again, these instructions take 3 operands, first is result,
the last may take immediate value. (2) Logical operations are BITWISE,
e.g.,
0101100...0110
and 1011001...1001
------------------------
result 0001000...0000
C
beq $2, $3, L "if($2==$3) goto L;"
...
L:
bne $2, $3, L "if($2!=$3) goto L;"
j L "goto L;"
jal f "f(a,b,c ...);"
slt slti sltu sltiu "$2<$3?"
Remark: (1) Conditional branch and jump instructions are used to control the execution of code, corresponding to C states like for(..), while(..), do {}, if(..) else. (2) jal is used for function call.
Move from memory to registers:
lw $2, address l.s $f0, address, l.d $f0, address
Move from registers to memory:
sw $2, address s.s $f0, address l.d $f0, address
Move between registers:
mflo $1
mfhi $1
mtc1 $1, $f0 (move from CPU to FPU)
mfc1 $1, $f0 (move from FPU to CPU)
Remarks: (1) Address is either of the form 100($4), for example, or a label (pseudo-instruction). (2) Only load/store instructions access memory. (3) A word occupies 4 bytes. (4) "Move" does not actually mean moving the data, "copy" is more appropriate.
move (copy contents) li (load immediate) li.s li.d (load immediate floating point) la (load address)
Remarks: (1) Pseudo-instructions (also called assembly instructions but not machine instructions) are translated into one or several machine instructions by the assembler. (2) Pseudo-instructions do not have machine encoding.
.data
str: .ascii "string\0"
A: .word 1
...
.text
myprog:
...
# code goes here
j $31
We should know these concepts. In particular, $2 is for integer type return value, $4 to $7 are for 1st to 4th integer type arguments.
All types of data on computer are ultimately represented by bit patterns. Given a bit pattern of 32 bits (1 word)
b_31 b_30 B_29 ...... b_3 b_2 b_1 b_0it can represent:
Remarks: (1) We should know how to negate a (two's complement) number, and sign extending a number. (2) Binary <-> decimal conversion, including floating point binary numbers. (3) Divide/multiply by 2 methods for binary/decimal conversions.
Following points are expected to know:
unsigned signed
1101 -> 13 -3 1101 -> -3
+ 0111 -> 7 7 x 0111 -> 7
------- -------------
(1)0100 -> 20 4 00000000
^ - 11111101
| ------------
overflow bit 00000011
+ 11101
------------
11101011 -> -21
Remarks: (1) Whether there is an overflow or not depends on the interpretation (signed or unsigned number). (2) Booth algorithm is for signed number. The result has twice the original bits.
(1) Read and understand the lecture notes. Read the relevant chapters of the textbook. (2) Do or re-do the tutorial/lab problems without referring to the answers. Then check against the answers. (3) If you have time or energy, do extra problems in the past exams, problems in textbook or other reference books.
Dr. Wang Jian-Sheng (office S16 02-35, tel x6880) is available for consultation in the reviewing period.