#byte access of FSR0H / FSR0L

[Torello]

Hi,

Does any body know why the compiler clears FSR0H when loading FSR0L ? First I though it might be a odd/even adress thing but if I do the same loading on for example ADCON1/2 then it does it OK...

[PCM programmer]

The quick solution is just to change the order of the lines.

[Torello]

I though of that solution too. But knowing -why- it is done give me more confidence that this solution stays working in the future....

A support request on this issue has been submitted to CCS.

(any thoughts are of course still welcome)

[Ttelmah]

I think you will find the reason is that the compiler already has a _16bit_ variable, called 'FSR0L' defined.

The #byte directive works in two different ways. If no variable exists with the name being used, it created an 8bit variable, and maps it to the defined address. If a variable already exists with the name, it maps this variable to the address.
So the compiler already has a 16bit variable called FSR0L defined, and when it sees the #byte directive, maps _this_ to the address. Hence when you access FSR0L, it accesses both addresses. Two solutions:

1) Use your own name to avoid any possible clash with the compiler 'My_FSR0L' for example.

2) Assume that the compiler's FSR0L, does not 'mind' being at the FSR address, and use:

[Torello]

Were is this _16bit_ applied on the definitions of Fsr's?

Solution 1 does -not- work: