| View previous topic :: View next topic |
| Author |
Message |
pic_user
Joined: 25 Aug 2005
Posts: 16
|
| Difference between V4.107 and V4.109 |
 Posted: Fri Jul 23, 2010 5:02 pm |
 |
|
I have successfully compiled a program in V4.107 but it fails in V4.109
The reduced version which shows the problem is:
| Code: |
#include <18lf26j11.h>
typedef struct menu_struct
{
ROM char *ASCII_ptr;
} menu_select;
menu_select test_ASCII;
void init_menu_structures()
{
#DEFINE MAX_MESSAGES 3
ROM char ccs_ASCII_mess[MAX_MESSAGES][20] = {"Hello", "goodbye"};
test_ASCII.ASCII_ptr = &ccs_ASCII_mess;
}
void main()
{
init_menu_structures();
while(1)
{
}
}
|
The complaint is about the assignment:
| Code: |
test_ASCII.ASCII_ptr = &ccs_ASCII_mess;
|
| Quote: |
*** Error 28 "ccs_data_structures.c" Line 48(37,51): Expecting an identifier
|
Has anyone else had this problem? |
|
 |
TimC
Joined: 12 Nov 2005
Posts: 7
|
|
| Posted: Fri Jul 23, 2010 9:05 pm |
|
|
It looks like ALL your variables and pointers are in ROM, that would be a neat trick if you could do it.
The following is from the manual:
Attempt to create a pointer to a constant
Constant tables are implemented as functions. Pointers cannot be created to functions. For example CHAR CONST MSG[9]={"HI THERE"}; is permitted, however you cannot use &MSG. You can only reference MSG with subscripts such as MSG[i] and in some function calls such as Printf and STRCPY.
I don't think V4.107 would work even though it compiled
Regards
TimC |
|
|
PCM programmer
Joined: 06 Sep 2003
Posts: 21708
|
|
| Posted: Fri Jul 23, 2010 9:16 pm |
|
|
If you want the address of an array, you don't put an '&' in front of it.
You just give it the name of the array, like this:
| Code: |
test_ASCII.ASCII_ptr = ccs_ASCII_mess;
|
Then it compiles with vs. 4.109.
Look in your .SYM file to see the address of 'ccs_ASCII_mess'.
Then use a printf statement to display 'test_ASCII.ASCII_ptr',
using "%lx", after the assignment is done. You'll see that it's
loaded with the ROM address of the array as given in the .SYM file.
(That is, after you remove the '&'). |
|
|
pic_user
Joined: 25 Aug 2005
Posts: 16
|
|
| Posted: Fri Jul 23, 2010 11:34 pm |
|
|
| I tried removing the & operator previously, as an experiment, and saw that it compiled without producing an error message. But when I used MPLAB SIM, I did not get the address in the ROM, where the message is located. Instead it was in RAM area. |
|
|
PCM programmer
Joined: 06 Sep 2003
Posts: 21708
|
|
| Posted: Sat Jul 24, 2010 12:32 am |
|
|
I couldn't test your code on a 18F26J11 on one PC here, so I tested the
18F25J10 and it worked. Now I switched to another PC and tested the
18F26J11 and it's not working. That's with the same basic program
too. It's getting late. I'll test this more on Sunday maybe. |
|
|
pic_user
Joined: 25 Aug 2005
Posts: 16
|
|
| Posted: Sat Jul 24, 2010 4:40 pm |
|
|
Thanks for the help.
Following your lead, I tried other devices, and sure enough they compile and simulate properly.
What is looks like is that the 64K ROM parts in the PIC18F46J11 are showing the problem - not the 16K or 32K parts.
It feels like a 16-bit address boundary problem.
I don't know what the workaround is, so I hope CCSINFO provides a fix.
As I said before, V4.107 works. The program fragment I posted was just to extract the isolated condition. In fact, the program is large and consists of multiple, complex structures. It works correctly. Only when I downloaded V4.109 did the problem show up. |
|
|
|
FAQ 
Search
Register
Profile 
Log in to check your private messages 
Log in
