Fault in CCS compiler version 3.136, HELP ME!!!!

[ivarboer]

Dear programmers,

Iam using a PIC16F877A processor in combination with CCS compiler version 3.316.

One of the code i use is the code below:

while ((Fl24HourRst) && (!FlWrtLastErrEep) && (!FlRunHigh1h) && (!FlRunLow1h) && (!FlBurnStrt));

Description:
Fl24HourRst = bit 7 of a byte (placed in bank 1 by the compiler)
FlWrtLastErrEep = bit 5 of a byte (placed in bank 0 by the compiler)
FlRunHigh1h = bit 0 of a byte (placed in bank 2 by the compiler)
FlRunLow1h = bit 1 of a byte (placed in bank 2 by the compiler)
FlBurnStrt = byte ((placed in bank 2 by the compiler)

Normally when Fl24HourRst = 1, FlWrtLastErrEep = 0, FlRunHigh1h = 0, FlRunLow1h = 0 and FlBurnStrt = 0 the programm has to stay inside the loop, but my programm doesn't.

I have checked the assemblycode and now i discovered that the compiler is not seting the right bank when jumping from the end of the while loop to the beginning of the while loop.

What is happening?
Isn't my code right?
Are there to many conditions in the while function?
Does the compiler contain errors?
Etc...

Iam using #device *=16

Please help me as soon as possible, because my EEPROM is overwritten sometimes.

Thanks!

Ivar

[neil]

Hi Ivar, there are rather a lot of conditions in that one expression, might be a case of function overloading but I think it will handle them. Clearly if the compiler is putting the flags in different banks, it makes for more instruction cycles because of all the bank switching.

Try putting your flags into a struct of bit variables, I think this will make the compiler assign the bits sequentially to one byte so no bank switching is necessary between parts in the expression. However, if you are doing this, it might be easier to apply a mask for the condition you are looking for, instead of bit testing each flag individually. apply to the whole byte and exor it. You may need to mask out the 'don't cares' with an AND mask.

eg.

struct flags{
boolean Fl24HourRst, FlWrtLastErrEep, FlRunHigh1h, FlRunLow1h, FlBurnStrt;
int unused:3;
} flag_byte;

Now, for the logic states in your expression apply a mask of: xxx10000. So to test the state:
do "flag_byte &= 0b00011111;
if((flag_byte ^ 0b00010000)==0){ do stuff }"

This will be true when the structure of flags matches your test mask. I think this will be more efficient code, but not as easily readable.

[PCM programmer]

If he uses the structure, he can also write the test like this,
and keep his existing readability.

[Ivar Boer]

Thanks for your information, but this information does not help me to determine the reason of my problem. Maybe the complier has made more wrong code which i don't know.

Could it be a compiler error????

Thanks.

Ivar

[SherpaDoug]

A compiler error is certainly a possibility. They are common in any uC compiler. But my first step would be to break that long line into smaller more digestable pieces. Then look at the assembler and see just what the compiler is doing. Then you will be able to find a way to get the behavior you want. The procedure is the same whether it is a compiler error or more likely a programmer error.
_________________
The search for better is endless. Instead simply find very good and get the job done.