Homework 4 Solution

Description

1. Consider the following code used to implement the instruction

.                sllv $s0, $s1, $s2

which uses the  least  significant  5 bits  of the  value  in register  $s2 to specify the  amount  register  $s1 should be shifted left:

.data

.text

mask: .word 0xfffff83f

start:  lw   $t0, mask     # load mask into $t0

lw   $s0, shifter  # load the instruction at label ‘‘shifter’’ into $s0 and  $s0, $s0, $t0 # mask out the shift amount (making it zero)

andi $s2, $s2, 0x1f # mask out the least significant 5 bits of $s2

sll  $s2, $s2, 6   # shift those bits left 6 places to align with shamt

or   $s0, $s0, $s2 # or the value from $s2 into the shift inst’n as shamt sw   $s0, shifter  # store the modified inst’n back where we got it

shifter: sll  $s0, $s1, 0   # execute the modified inst’n

Add  comments  to  the  code and  write  a paragraph describing  how it  works.   Note  that the  two lw instructions are  pseudoinstructions that  use a label  to  specify a memory  address  that contains  the word of data  to be loaded.  Why do you suppose that writing  self-modifying code such as this is a bad idea (and  oftentimes  not actually  allowed)?

2. The following MIPS  instruction sequence could be used to implement a new instruction that has two register  operands.   Give the  instruction a name  and  describe  what  it does.  Note  that register  $t0 is being used as a temporary:

srl  $s1, $s1, 1  #

sll  $t0, $s0, 31 # These 4 instructions accomplish srl  $s0, $s0, 1  # new ‘‘$s0 $s1’’

or   $s1, $s1, $t0 #

3. Write a program  in MIPS  assembly  language  to convert  an ASCII decimal string  to an integer.  Your program  should  expect  register  $a0 to  hold the  address  of a null-terminated string  containing  some combination of the  digits  0 through 9.  Your program  should compute  the  integer  value equivalent to this  string  of digits,  then  place the  number  in register  $v0.  Your  program  need not  handle  negative numbers  and need not be concerned  about  values larger than  231  − 1.

If a nondigit  character appears  anywhere  in the  string,  your  program  should  stop  with  the  value  -1 in register  $v0.  For  example,  if register  $a0 points  to  a sequence  of three  bytes  5010 , 5210 , 010   (the null-terminated ASCII string  24), then  when the program  stops,  register  $v0 should contain  the value

2410 , (or 11000 in binary). (The  subscript ten means base 10.)

Submit your solution to question 3 on  WebCT.

1