Microprocessors Lecture 4

Addressing Modes

It is important to remember that although we are looking in detail at the M6809 we need to learn more general lessons about the way in which such microprocessors are organised. Thus the addressing modes we are looking at will appear in one form or another in all processors but may have different names or may work in a slightly different way.

Indexed Addressing

This is a mode which at first appears unnecessarily complicated but in fact turns out to be very useful, particularly when tables of data have to be accessed. It is also the only mode which allows addresses to be calculated at run time, i.e. it must be used where the address is not known when the program is written.

The basis is that the address is held in one of the index registers in the processor, X or Y. Obviously it must be put there in previous instructions. This base address is then modified by an offset which is included in the instruction. The offset is interpreted as a twos complement number , which means that it can be both positive and negative depending upon the most significant bit.

2's Complement Numbers (digression)

There are many ways of looking at two's complement numbers - the simplest is to regard it as shifting the starting point for a set of numbers.
                    Normal                        Twos Complement
                Binary      Decimal              Binary      Decimal
               00000000         0               1000000       -128
               00000001         1               1000001       -127
		   .				    .
		   .				    .
               01111111         127             1111111         -1
               10000000         128             0000000         0
               10000001         129             0000001         1
		   .				    .
		   .				    .
               11111110         254             0111110         126
               11111111         255             0111111         127
Note that for positive numbers 0 to 127 the binary values in both systems are the same. If we want to find out what a negative binary two's complement number 'means' we invert all the bits and add 1.

E.g. 10110101 = minus (01001010 + 1) = minus 01001011 = -75

Fortunately the same process works the other way if we wish to convert a negative number into two's-complement notation.

E.g. -25 = minus 00011001 = 11100110 + 1 = 11100111

Here we have been using 8 bits but it works equally well with any number of bits e.g. 16 or 5

Using 16 bits minus 0000000000011001 = 1111111111100110 + 1 =1111111111100111

Using 5 bits minus 11001 = 00110 + 1 = 00111

But something has gone wrong here! 00111 is 7 in binary and not a negative number at all. In other words, 25 or -25 is too big to represent using 5 bits.

                10000      -16
                10001      -15
                10010      -14
		 ...
		11111	   -1
                00000       0
                00001       1
		 ...
                01111       15
To return to indexed addressing, the actual or effective address is made up in the processor by adding the contents of the index register and the offset contained in the instruction. The actual instruction format, however, is complicated because these are some of the extended instructions referred to earlier.
                Load Acc A with an offset of 5 from Y    LDA 5,Y
The basic code for 'Load Acc A using indexed addressing' is A6. The next byte contains two pieces of information. Firstly that IRY is being used and secondly that the offset is 5. The format of this byte is
                              ORRnnnnn

           where RR is a code representing the register
                      00 - Index register X
                      01 - Index register Y
                      10 - User stack pointer U
                      11 - System stack pointer S

           and nnnnn is a five bit offset as detailed above.
The second byte is therefore 0 01 00101 or 25 and the complete instruction code is therefore A6 25

Notice that the stack pointers (S, U) can be used in a similar way to the index registers.

Obviously we are frequently going to need a larger offset than can be accommodated in a 5-bit offset. We can in fact use an 8 or 16 bit offset but the instruction format must change because we can't fit all the information required into a single byte. If the first bit of the second byte is a 1 then it indicates to the processor that a different mode is going to be used. The format becomes

                         1RRccccc
where RR is the same as before indicating which address register is to be the base.
and ccccc is an additional instruction code as follows
              00100 - No offset (has the same effect as 00000 in 5 bit)
              01000 - 8-bit offset in next byte
              01001 - 16-bit offset in next two bytes
Obviously we could have 3 different instructions all having the same effect.
              A6         A6         A6
              25         A8         A9
                         05         00
                                    05
Clearly the first is going to be preferable on the grounds that it takes up less memory and executes more quickly since there are fewer bytes to be read.

What about all the other possible codes? Each of them in fact allows an additional mode of addressing which we will not describe in detail, but these are just list listed.

  1. Auto increment/decrement
    	e.g. 	LDA ,X+
    		LDD ,X++
    		LDA ,-Y
    		LDX ,--Y
    
                00000 - Use no offset but increment the address register by 1
                00001 -     no offset but increment the address register by 2
                00010 -     no offset but decrement the address register by 1
                00011 -     no offset but decrement the address register by 2
    
  2. Accumulator offset
    	e.g.	LDA B,Y
    		LDB A,X
    		LDA D,X
    
                00110  - Use 8-bit offset in Accumulator A
                00101  -     8-bit offset in Accumulator B
                01011  -    16-bit offset in Accumulator D (A,B)
    
  3. Relative
    	e.g.	LDA 8,PC
    		LDB $1000,PC
    
                01100  - Use 8-bit offset from the Program Counter (RR ignored)
                01101  -    16-bit offset from the Program Counter
    

Indirect Addressing

If the first bit of ccccc is a 1 then the addressing mode becomes indirect. This means that the Effective Address is calculated by the processor and the contents of that address (and the one following) is used to form a second address. The second address is where the data is stored. Note that this requires several memory accesses; two accesses to retrieve the 16-bit address and a further access (or accesses) to retrieve the data which is to be loaded into the register.

N.B. It is not essential to learn Indirect Addressing for this course

Relative Addressing

One of the problems encountered early on in programming is that if you write a program using addressing modes such as direct and extended, then if you move that program somewhere else in memory then it no longer works. Why should you want to move the program somewhere else? The most obvious reason is that you want to build up a library of useful bits of program or subroutines so that you can use them each time you're writing some software and so don't have to keep on rewriting the same bits of program. Obviously there is no guarantee that they will end up in the same place in memory. Another reason is that in a multi-user environment many users, programs will be loaded into memory at once. Again there is no guarantee that the same area of memory will always be available.

Relative addressing is one of the techniques which alleviates this problem because using this mode addresses are relative to the program counter i.e. the address of the instruction. Thus

           Load Acc A with contents of PC+128	LDA 128,PC
will work correctly wherever the program happens to be.

As we have already seen we can use relative addressing for data movement instructions but the chief use comes in branch instructions. Branch instructions are used to create loops in a program and so are frequently of the form

            Branch back 30 bytes if the last arithmetic instruction 
	    resulted in zero.                   BEQ PC-30
Most loops are relatively small and within range of an 8 bit twos complement offset. All the branch instructions in the 6809 therefore use an 8-bit offset relative addressing mode. The above instruction becomes
                                   27
                                   E2    -30 dec
Where an offset becomes too large for this mode then a long relative mode is available in which the code is prefixed by 10
                                   10
                                   27
                                   FF
                                   E2
This is an equivalent instruction using 16 bit offset
Similar instructions using alternative addressing modes, direct and extended are provided and for some obscure reason these are called jumps rather than branches.

Clearly there are a lot of powerful addressing modes here, some of which are not easy to understand. The important ones to remember are

Condition Code or Status Register

We have examined the role of most of the registers in the 6809, the ones remaining are the U and S stack pointers and the condition code register. The latter is not really a single register at all but is made up of 8 single bit registers. These 8 bits are labelled (in order) E F H I N Z V C and are used by the processor to store the fact that certain events have occurred. These events fall into two categories, those resulting from arithmetic operations and those resulting from external interrupts.

Arithmetic Status Bits

          C   Carry/Borrow     When two 8 bit numbers are added a 9 bit
                               number may result.  The 9th bit is transferred
                               to the C bit.
                                          F3              F3
                                     +    0E         +    07
                                     =  1 01         =  0 FA

                               Similarly in subtraction, if a larger number
                               is subtracted from a smaller then a Borrow
                               results which is transferred to the C bit.
                                          56              66
                                     -    63         -    63
                                     =  1 F3          = 0 03

          N   Negative         In twos-complement notation the most
                               significant bit can be regarded as the sign
                               bit - if it is 1 the number is negative.  This
                               sign bit is transferred to the N bit so that
                               it can be used in Branch instructions.

          Z - Zero             The Zero bit is set to 1 whenever the result
                               of an arithmetic operation is zero.  The Zero
                               bit is cleared whenever the result is
                               anything but zero.

          V - Overflow         This is similar to the C bit but relates to
                               twos-complement arithmetic.  When two positive
                               numbers are added together they can give a
                               negative result without causing a carry.  The
                               V bit is set when this situation occurs.
                                          73
                                      +   24
                                      =   97   = -69

          H - Half Carry       In BCD notation four bits are used to
                               represent a decimal number with the values A
                               to F not allowed.  When two such numbers are
                               added the result may have to be adjusted. to
                               correct for the unallowed digits.

                                          13            19
                                     +    85       +    85
                                     =    98       =    9E - needs adjusting

                              The adjustment is carried out by the Decimal
                              Adjust instruction as follows. 06 is added
                              and the H bit checked.  If it is 1 the
                              adjustment was needed otherwise the 06 is
                              subtracted again. 98 + 06 = 9D (no adjustment).

                                     9E + 06 = A4 (adjustment needed)

                              The process is then repeated with 60 and
                              checking the Carry bit.

                                     98 + 60 = F8   (no adjustment)
                                     A4 + 60 = 1 04 (adjustment needed)

Interrupt Status Bits

          I - Interrupt       
          F - Fast Interrupt 
          E - Entire
These are all associated with interrupt handling which we will deal with later.

Flags

With the exception of the Half-carry bit these are used in branch instructions in various combinations or individually.
          BCC - Branch if Carry Clear (C=0)
          BCS - Branch if Carry Set    (C=1)
          BNE - Branch if Not Equal to zero (Z=0)
          BEQ - Branch if Equal to zero (Z=1)
          BLS - Branch if lower or same (C+Z=1)      This is used after a
          compare instruction and results in a branch if the first number
          is lower (C=1) or the same (Z=1) as the second.
          etc.

| Back | Next |