How many of you are having ram allocation problems?

[future]

Hi,

I would like to know how many of you, like me, start having problems with code that was previously working before the project gets bigger.

I have functions that worked fine for years and simple code addition to other sections makes them fail.

It seems that the compiler optimization is badly reusing its ram space.

The compiler version is 3.249 with a 18f452.

[PCM programmer]

Have you looked at the Symbol Table ? Can you see if some RAM
locations are re-used when they should not be ?

[future]

Yes, there is a major reusage here... but I don't know which one would be correct.

Lowering the optimization level does not seem to solve the problem.

[PCM programmer]

It's normal for CCS to re-use RAM for local variables after their
existance has expired -- for example, when a function exits and
returns control back to main().

One place where it would a problem, would be if an interrupt routine
used RAM for local variables that was used by other routines in the
program. From just the little code that you posted, I can't really tell
if this is happening.

Some of your symbols are called "Task". Does this mean you're
using the CCS RTOS ? I don't have that, because I only have the
command line compilers.

[future]

Yes, I am using the RTOS only to time these tasks.

I wonder how to trust the code having problems like this... How to know if another locations aren't being changed?

You seem to have a good experience with firmware writing. What kind of debugging tools do you use?

The icd gives me a great help while writing routines alone, but it is very hard to debug a more complex system. I have passed the 11k lines barrier and it is a headache that a already tested module starts failing.

[PCM programmer]

The only time we use a commerical RTOS here at the company is for a
large project where the customer is willing to pay for it. Then, the
software guy here does all the coding. I'm not the one who does it.
I do the smaller projects that don't require a commerical RTOS.

[Ttelmah]

Let me give you an example of the sort of thing that can cause problems.
The first would be, a subroutine that returns a pointer to a local variable. It will initially work, but will fail once RAM useage leads to re-use. Now this is not the compilers 'fault', but a failure in programming, the variable should be declared as static to prevent this.
A second example, would be a local routine, that 'assumes' a local variable is zero when it is called. Relying on 'zero RAM' to zero it in the first cal, and then clearing it itself before exiting. Again this will initially work, but will fail, if something else uses the area. Again the problem is the result of careless programming.
Key rules that avoid these problems are:
1) Always initialise variables on entry to routines.
2) Use static variables for any data accessed using pointers.

Now these are just two examples, but most problems of this sort, come from the same type of sources.

Best Wishes

[future]

It seems that the compiler is having problems with temporary storage. If I simplify the code breaking big expressions in shorter ones the problem goes away.

I'll investigate what is exactly causing the problem.

Thank you for the tips.