CSE3101 Fall 2004 Exam #1, open book, open notes.  Name ___________________

1.  What is the difference between real mode and protected mode? (10 pts)

  ANSWER: Real mode has a 1 MB address space, 16-bit pointers, and full
  access to all instructions.  It is intended for single-tasking operating
  systems.  Protected mode has a 4 GB address space and 32-bit pointers.
  Programs are assignmed memory segments with read, write, and execute
  permissions.  Access outside these segments is trapped by the operating
  system.  Only the operating system may modify assigned segments using
  priveliged instructions.

2.  Which mode does Windows use? (5 pts)  ANSWER: protected

3.  Which are valid MASM statements? Circle Yes or No (2 pts each).

                                     ANSWERS
  x db 10 dup(?)                     Yes
  y dd 1,2,3                         Yes
  mov x, y                           No, can't address 2 memory locations
  mov 20, al                         No, move is right to left
  mov [esi+ebx*8+offset x], al       Need "mov byte ptr [esi..." (I accepted either answer)

4.  Translate the following C function to MASM (20 pts).

  int f(int a, int b) {return a+b;}

  ; ANSWER
  public _f         ; needed if called from C
  _f:
  mov eax, [esp+4]  ; a
  add eax, [esp+8]  ; b
  ret               ; result returned in eax


5.  Write MASM equivalent to the following C code.  Assume x and y
are arrays as in question 3 (25 pts).

  y[1] = f(3, (x[5] << 4) - 8);

  ; ANSWER
  mov eax, [x+5]    ; gets 4 bytes, x[5]...x[8]
  and eax, 0ffh     ; don't need x[6]...x[8] so set to 0
  shl eax, 4        ; x[5] << 4
  sub eax, 8        ; (x[5] << 4) - 8
  push eax          ; pass arguments right to left
  push 3
  call _f
  add esp, 8        ; In C, caller pops arguments off stack
  mov [y+4], eax    ; result from f


6.  How many times does this loop? (10 pts)

  L1: mov ecx, 10h        ANSWER: forever (note position of label)
      dec ecx
      jnz L1

7. If a, b, and c are 32-bit unsigned numbers (defined with dd in MASM),
write MASM code equivalent to a = a*b/c; such that the 64-bit intermediate
result a*b does not overflow.  You may assume c > b (10 pts).

  ; ANSWER
  mov eax, [a]
  mul [b]           ; a*b in edx:eax as 64 bit number
  div [c]           ; divide edx:eax / c, put quotient in eax, remainder in edx
  mov [a], eax      ; don't forget to put result in a


8.  Assume a and b are 64-bit unsigned numbers in little-endian format
(defined with dq).  Put a+b into edx:eax (10 pts).

  ; ANSWER
  mov eax, [a]      ; low 32 bits of a
  mov edx, [a+4]    ; high 32 bits
  add eax, [b]      ; add low 32 bits of b
  adc edx, [b+4]    ; add high 32 bits + carry