Just one change to the bootloader example....

RckRllRfg · Joined: 20 Jul 2011 Posts: 60

...and it seems to go astray.

I have spent a substantial amount of time on this (over 40 hours since last Saturday) and I need to get things moving forward. Any help is truly appreciated. I've literally picked apart the example bootloader code, I have read numerous forum postings, and I have RTFM when I didn't understand a command. I have printed and read PCM Programmer's outstanding post on how to adjust the memory allocations:

http://www.ccsinfo.com/forum/viewtopic.php?t=45044&highlight=ive+created+technique+solve+problem

What is going to drive me to drink is the memory allocation. At one point, I had coaxed it to get me past the infamous:

"Out of ROM. A segment or the program is too large real_load_program SEG 0004A-001FF,0169 left, need 001CC"

To keep a long story short, I found that I was doing a stomp job on the loader.

So, I took a step back and started with a clean copy of the bootloader example. I trimmed out all of the PCH options and I added a few output_high and output_low commands to handle the transceivers. I have to use UART2, which has not been an issue.

I still got the error message.

I made a worksheet that follows PCM programmer's outline for the memory allocation. Attempted to move all of the locations, widen locations, etc.

Still get the same error.

...At this point, make mine a scotch, please!

So, I went to the original bootloader code with NO changes what so ever. I only change the header file. Got the same error message.

If any one wants to give it the old college try or has recommendations, I would truly appreciate it. The only thing that needs to be changed in the bootloader code is the header file, from <16F877.h> to <16F1947.h> and voila, instance error message!

Using a PIC16F1947 (you don't have to have one...I just need someone to successfully compiler it)

Using version 4.123 with MPLAB 8.73.

Thanks for your help and understanding,

RckRllRfg

FvM · Joined: 27 Aug 2008 Posts: 2337 Location: Germany

The answer is simple. The 16F1947, which is called Enhanced MID Range CPU implements additional registers and increases some registers to 16 Bit. As a result, the code size increases. The bootloader as is won't fit in the 0..1FF segment any more, it has to be increased by at least 2 rows to 0..27F.

Unfortunately, bootloader protection is only available for 0..1FF or 0..FFF, so it would be preferably to shrink the bootloader.

However, if you don't rely on bootloader protection, adjusting the limits is rather easy.

RckRllRfg · Joined: 20 Jul 2011 Posts: 60

Hi FvM

Thank you for your reply.

Unfortunately, I do not have a choice when it comes to shrinking the bootloader. In fact, I have to add to it in order control my transceivers. Fortunately, my main application uses about 40% of the memory, so I have plenty of room for the bootloader.

I have been adjusting the LOADER_END,LOADER_SIZE, and the main(); however, other than pushing the main() backwards to #0RG 0x11, 0x3F, I have not be successful in adjusting the widths or lengths of the memory space.

Is there a protected region? Is there a command that I can declare that turns off the protection? I'm also thinking about the use of the #Seperate and possibly splitting up the real_load_program.

Thanks for your insights,

RckRllRfg

FvM · Joined: 27 Aug 2008 Posts: 2337 Location: Germany

I tried my example with V4.112, which is my working CCS C version, and just noticed, that the size of the main() stub has increased a few bytes with recent versions. With V4.124, I corrected the settings to

RckRllRfg · Joined: 20 Jul 2011 Posts: 60

Just to confirm: You use CCS version V4.112 with the PIC16F1947.h file substituted into the example bootloader code and got it work? It this correct?

The CCS website is vague on details when it comes to releases. Do you know approximately when V4.112 was released?

Thanks -

RckrRllRfg

RckRllRfg · Joined: 20 Jul 2011 Posts: 60

I did try the values as FvM provided:

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

It compiles for me. I installed vs. 4.112. I made a project in the
Examples directory using the CCS Ex_bootloader.c program as the
source file for the project. I made the suggested changes as shown in
bold in the two files below. Result:

FvM · Joined: 27 Aug 2008 Posts: 2337 Location: Germany

V4.112 was released about one year ago. It seems to support 16F1947, but I don't know if there are any issues. However, as I said, I used recent V4.124 with the final settings. They work with the original ex_bootloader.c example, the said changes apply to bootloader.h and ex_bootloader.c. There's also a LOADER_SIZE definition in loader.c that should be probably adjusted, but it doesn't cause segment errors.

You get "Invalid org range" with #org 0x20,0x47 if you didn't modify LOADER_END and LOADER_SIZE in bootloader.h, as suggested.

RckRllRfg · Joined: 20 Jul 2011 Posts: 60

FvM and PCM Programmer -

Thank you for the responses. After I had posted, I was staring at the error message. It was driving me crazy that the values were not changing. (ie, 40-49 was ALWAYS the space for the load_program function.)

Sure enough, I remembered that when I first went about doing this several days ago, I had copied and changed the name of the bootloader.h and loader.c to avoid any ambiguity with the original files. This remedied the issue at the time. In the mix of it all, I had forgotten all about it.

As mentioned earlier, I decided to start fresh with the example code and changed only that what I needed. My original code is fine; however, I added a TON of comments in it when I was first reading through it to understand its mechanics. I also added many print statements. I was seeing memory stomping and thus decided to step back with a clean example. That is when I started stumbling over the memory issue again.

I was able to get it to compile. One minor tweak was needed for the main. I am jazzed about that! Thank you for guiding me along.

Apparently, the compiler is looking at the example directory in the C:\Program File\PICC installation directory for the bootloader.h. That's a bit scary! (and confirms FvM's suspicion that bootloader.h was never being modified)

BUT, I'm not completely out of the woods just yet....

The issue that I was seeing with my original code seems to have resurfaced. My memory allocation for the loader ends at 3FF. Therefore, the application should start at 400h and go on from there; however, I am seeing code from the bootloader being placed in the 400h addresses.

Any thoughts on that?

Again, many thanks for getting me through thus far.

RckRllRfg

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

FvM · Joined: 27 Aug 2008 Posts: 2337 Location: Germany

Yes, directory usage of CCS C is confusing, particularly if you have multiple versions installed in parallel. At least when I modify CCS C provided files, I prefer to have a copy in the application directory, preferably with a different name.

RckRllRfg · Joined: 20 Jul 2011 Posts: 60

PCM Programmer -

I am using the following memory allocations:

#org 0x20, 0x7A

#define LOADER_END (0x3FF)
#define LOADER_SIZE (0x37F)

I also widened some of the other functions. necessary? Not really; however, did it to ensure that everything fit. I can tighten this up later.

Just before the void real_load_program in the loader.c (approx line 46)

RckRllRfg · Joined: 20 Jul 2011 Posts: 60

This is the code that I am seeing in the .lst file:

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

I don't get this. You're constantly changing things. We're always one
step behind you. We fix what you currently say. But by then you've
changed it and the new code doesn't work...

What is this ? A custom file ?

RckRllRfg · Joined: 20 Jul 2011 Posts: 60

Hi PCM programmer -

I agree that I am bouncing about here. I'm trying things on my end while getting responses to you. For that, I apologize.

The bootloader name was changed to ensure that my compiler was not going off and looking for it in the Program Files directory. It contains the addresses I mentioned above.

Here's my bootloader file. I stripped it of the PCH stuff: