write_virtual_eeprom() and ROM/RA usage: How to minimize?

[apakSeO]

CCS v5.073
PIC16F1503

Is there a way to lessen the amount of program space and/or RAM the write_virtual_eeprom() function is using? The PIC in my application is very limited with 2K program space and only 128 bytes of RAM. I've been pulling hairs trying to limit the code and variable usage and I think I am at the practical limit of what I can acheive.

In my application code, prior to inserting the line

[Ttelmah]

In these chips, a 'routine', must be small enough to fit inside the page. Your chip is not out of ROM, but you have a routine (your main), which is a too large 'lump'. You need to split it up. A search here for:
"Out of ROM, A segment or the program is too large"
Will find dozens of threads about this descibing how to split the code up.

Writing to the program memory is quite complex. Unfortunately you are meeting a limitation of your chip. This is why people will say if you want eeprom, get a chip with eeprom...

If you only want (perhaps) a maximum of 16 bytes, then use an alternative approach.

Have a RAM array of 32bytes.
Read a single 32byte page of the flash using read_program_memory.
Use this for your config data, only using the low byte of each 16bits.
Make any changes you want, and write the entire 32bytes back.

You have to write a whole page, since this is the block that will erase. when a write is done.

[apakSeO]

I read as many threads here in 3 hours I could find regarding this "Out of ROM" message & the #separate directive. Unfortunately, this doesn't seem to do anything at all for me when I implement it in source code. I do not see a change in the ROM usage at all.

I was able to refactor and gain a little bit of ROM space by consolidating functions into a single function, and that allowed me to call write_virtual_eeprom() a single time, without compile errors. Calling it a second time leads to the Out of ROM error.

I've even tried to delve into the virtual_eeprom.c driver and put #separate before each function definition. I can't tell if this helped matters at all, for I do not see a change in the ROM usage. Without the any call to write_virtual_eeprom(), my Memory usage: ROM=83% RAM=31%-85%

Including a single call to write_virtual_eeprom() within an if {} block jumps the Memory usage: ROM=97% RAM=31%-95% A second call to write_virtual_eeprom() takes me over the edge and the program will not compile, resulting in:

[PCM programmer]

Order the 16F1825, put your code in that with read_eeprom() and
write_eeprom() and you're done.

You could do a test compile of that now. It would brighten you up since you
would see a solution.

[apakSeO]

Thanks for the kind sentiments, PCM

In order to fit the 16F1503 on this tiny PCB, I had to go with the 3mm x 3mm UQFN The 16F1825 does not offer this package option, regrettably, and is only offered in the 4mm x 4mm UQFN as far as I can tell.

I was hoping I could replace the 16F1503 on these PCBs, for I am more or less a PCB rework expert, but alas, seems not to be the case. In rev 2, I'll have to move onto the 1825.

If I figure this out, I'll reply back here for posterity.

[Ttelmah]

As a comment though, you don't want to consolidate, you want to _separate_. Change the declaration to write_virtual_eeprom. Declare it as #separate. By default the compiler tends to 'inline' code. Do the same to any other of your own functions you call. You may well then find things fit.