Transcript 16-Parameter_Passing

© 2010 Kettering University, All rights reserved.
Microcomputers I – CE 320
Jaerock Kwon, Ph.D.
Electrical and Computer Engineering
Kettering University
[email protected]
http://www.kettering.edu/~jkwon
Announcements
• Midterm result




Average: 68.81
Standard Dev: 13.25
Maximum: 95
Minimum: 41
• Update grades on Blackboard
 Q1, Q2, Q3, and Midterm
 Lab1, Lab2, Lab3, and Practical exam
• Question 1, 2, and 3 in Lab2




I intended to have actual assembly code.
Questions were not clearly specified.
I will accept proper explanations as a right answer.
Feel free to stop by my office after this class.
Midterm Review
• 1. Branch instructions
 BHI (unsigned), BGT (signed)
• Take a look at the preceding comparison instruction.
• Then, you can use this instead of using complex formula in the
instruction reference.
 BRANCH IF REGISTER IS HIGHER/GREATHER/… THAN OPERAND
• e.g. CMPA #$D0
 Branch if ‘A’ is HIgher than ‘operand.’
 BMI, BVC, BNE
• Check the result if it is Minus/signed oVerflow/Not Equal to zero
 CMPA: A – (M)
 TSTA: A – 0
 V bit (signed overflow)
•
•
•
•
N+NP
P+PN
N–PP
P–NN
Midterm Review – cont’d
• 2. BRA and LBRA
 Write a line of assembly code that begins in memory location
$2450.
 BRA <8-bit signed offset value>
 LBRA <16-bit signed offset value>
 When the effective (destination) address is calculated in relative
addressing mode, a value that you need to add to an offset is PC.
 We have to use PC which is the next address of the line of
assembly code of (L)BRA
Midterm Review
• 3. Mostly this is about calculating postbyte(s)
 -20: You cannot use 5-bit offset. You have to use 9-bit one.
Lecture 16:
Parameter Passing
Today’s Goals
• Parameter passing
• Understand how to pass parameters using the stack
Call by Value vs. Call by Reference
Value
Subroutine
Main
Program
Memory
Address
Subroutine
Look for
values
Passing Parameters in Registers
• Use registers to pass parameters
• The simplest type of parameter passing
• Pros vs. Cons





Data is immediately available to the subroutine
No extra preparations are needed.
Often fastest execution and smallest code size
Only a limited number of registers are available.
We can use ‘pass-by-reference’ to point a list of input/output
values.
Example (Parameter Passing in Registers)
• Requirements
 A subroutine that adds an array of two-byte numbers and a
sample main program that calls it.
 The array is passed by reference in X.
 The length is passed by value in D.
 The sum should be returned by value in D.
 Do not worry about indicating signed or unsigned overflow.
Example (Parameter Passing in Registers)
Array
Length
ORG
dc.w
dc.w
$3000
$1234,$5678,$ABCD
3
ORG
LDS
LDX
LDD
JSR
SWI
$2000
#$3600
#Array
; load X with address of list
Length
; Load D with actual length
sumword
sumword TFR
LDDD
CPY
Loop
BEQ
ADDD
INX
INX
DEY
BRA
endsum
RTS
D,Y
#0
#0
endsum
0,X
loop
Preserving Registers
• In the previous example, the subroutine uses Y. The value in
Y is destroyed.
• If Y was in use by the main program, the main program would
continue after the subroutine with an incorrect value in Y.
• To avoid this, registers used by a subroutine may be saved to
the stack before a subroutine uses them.
• The saved registers are restored after the subroutine.
• We have two options
 The caller (main program) does this.
 The callee (subroutine) does this.
Preserving Registers
• Responsibility of the Caller:
 The calling main program assumes all registers are destroyed by
the subroutine
 If the registers used by the subroutine are unknown (i.e. using a
sub. provided by someone else), may save registers that the
subroutine wouldn’t affect
 Code to save/restore registers duplicated with every subroutine
call.
• Responsibility of the Callee:
 Saves only the registers that will be used by the subroutine
 Save/restore code only occurs once
Example – Caller Responsible
Array
Length
JSR
ORG
dc.w
dc.w
$3000
$1234,$5678,$ABCD
3
ORG
LDS
LDX
LDD
PSHX
sumword
PULX
SWI
$2000
#$3600
#Array
Length
sumword TFR
LDDD
CPY
loop
BEQ
ADDD
INX
INX
DEY
BRA
endsum
RTS
D,Y
#0
#0
endsum
0,X
; load X with address of list
; Load D with actual length
Stack Frame
RAH
RAL
XH
loop
XL
Example – Callee Responsible
Array
Length
sumword
TFR
loop
endsum
RTS
ORG
dc.w
dc.w
ORG
LDS
LDX
LDD
JSR
SWI
PSHX
PSHY
D,Y
LDDD
CPY
BEQ
ADDD
INX
INX
DEY
BRA
PULY
PULX
$3000
$1234,$5678,$ABCD
3
$2000
#$3600
#Array
; load X with address of list
Length
; Load D with actual length
sumword
#0
#0
endsum
0,X
Why not D?
It is used to return the
answer!
Stack Frame
YH
YL
XH
loop
XL
RAH
RAL
Passing Parameters in the Stack
• Pros and Cons




The stack pointer is already in use for the return address
Indexed addressing can easily access data stored on the stack
The amount of data passed is not limited by the register set
The data passed on the stack must be removed, and this is
usually the responsibility of the caller (although this can be done
by the callee)
• Note:
 If the caller will save registers on the stack, it should be
done before passing parameters.
 Why? Those parameters are supposed to be used in the
subroutine.
Example – Param Passing in the Stack
ORG
dc.w
dc.w
$3000
$1234,$5678,$ABCD
3
ORG
LDS
LDX
PSHX
LDD
PSHD
LDD
PSHD
JSR
LEAS
PULX
SWI
$2000
#$3600
#$1234
sumword LDDD
LDX
LDY
Loop
BEQ
ADDD
INX
INX
DEY
BRA
endsum
RTS
#0
2,SP
4,SP
endsum
0,X
Array
Length
; something to save
Length
#Array
sumword
4,SP
Stack Frame
RetH
RetL
ArrH
ArrL
LenH
LenL
XH
loop
XL
Questions?
Wrap-up
What we’ve learned
• Parameter Passing
• Pros and Cons of each method
What to Come
• More about subroutines