 |
 |
| View previous topic :: View next topic |
| Author |
Message |
Pekub
Joined: 20 Dec 2005
Posts: 16
|
| Pointer to struct V4.127 |
 Posted: Tue Nov 22, 2011 5:56 am |
 |
|
Hello,
To access the elements of my structure I used the pointer. It worked fine with Compiler Version 4.122 and older.
With new version V4.126 and V4.127 it doesn`t work.
I wrote simple program to demonstrate it. See variables b0 and b1 in Func().
| Code: |
#case
#include <18F26K22.h>
#FUSES INTRC //Internal RC Osc
#use delay(clock=16000000)
struct AA
{
int8 x;
int8 y;
int8 Buff[16]; // prijimaci buffer
};
struct AA Probe;
void Func(struct AA *pRx)
{
int8 *ptr;
int8 xx, yy, b0, b1;
ptr = &pRx->x; // pointer to Probe.x
xx = *ptr; // x = 5 works o.k.
ptr = &pRx->y; // pointer to Probe.y
yy = *ptr; // y = 6 works o.k.
ptr = pRx->Buff+0; // pointer to Probe.Buff[0]
b0 = *ptr; // b0 = 3 with V4.122
// b0 = 0 with V4.127 !!!!!!!!!!!!!!!
ptr = pRx->Buff+1; // pointer to Probe.Buff[1]
b1 = *ptr; // b1 = 4 with V4.122
// b1 = 0 with V4.127 !!!!!!!!!!!!!!
}
#zero_ram
void main()
{
Probe.x = 5;
Probe.y = 6;
Probe.Buff[0] = 3;
Probe.Buff[1] = 4;
Func(&Probe);
while (TRUE)
{
}
}
|
What is wrong ?
Thank`s
Peter |
|
 |
PCM programmer
Joined: 06 Sep 2003
Posts: 21708
|
|
| Posted: Tue Nov 22, 2011 2:30 pm |
|
|
If I cast the pointers to int8* before using them, then it works. This is a
fairly common type of "fix" that has to be done with CCS, sometimes.
The program below has this output in MPLAB simulator:
| Code: |
#include <18F26K22.h>
#fuses INTRC_IO
#use delay(clock=16M)
#use rs232(baud=9600, xmit=PIN_C6, rcv=PIN_C7, ERRORS)
struct AA
{
int8 x;
int8 y;
int8 Buff[16]; // prijimaci buffer
};
struct AA Probe;
void Func(struct AA *pRx)
{
int8 *ptr;
int8 b0, b1;
ptr = (int8 *)(pRx->Buff) +0; // Cast pointer to int8*
b0 = *ptr;
printf("b0 = %x \r", b0);
ptr = (int8 *)(pRx->Buff) +1; // Cast pointer to int8*
b1 = *ptr;
printf("b1 = %x \r", b1);
}
//===================================
void main()
{
Probe.x = 5;
Probe.y = 6;
Probe.Buff[0] = 3;
Probe.Buff[1] = 4;
Func(&Probe);
while (TRUE)
{
}
} |
|
|
|
hmmpic
Joined: 09 Mar 2010
Posts: 314
Location: Denmark
|
|
| Posted: Tue Nov 22, 2011 3:18 pm |
|
|
Hi
Please report it to CCS.
I have reported some other error on 4124->4127.
I think there came more and more funny errors every time there are a new release.
A major problem is to know what there are really changed, because all the new problem I not stated in the change log?
Sorry to say it:-( |
|
|
Ttelmah
Joined: 11 Mar 2010
Posts: 19552
|
|
| Posted: Tue Nov 22, 2011 3:51 pm |
|
|
This is why 'old users' of CCS, have a number of mantras:
1) Don't treat the releases as useable, till _you_ have tested them. CCS 'releases', are at best 'beta' code.
2) If writing a program, and you don't _need_ a fix that supposedly comes with a new 'release', stick with the older compiler.
3) When you have written code with a particular compiler version, save both the code, _and the compiler used_ together.
This is why people like PCM_Programmer, have access to dozens of older versions. I've got at least a dozen V3 compilers, and even some V2 ones, still stored, and the source code for my programs, always has at the top, just after the fuses etc., details of what compiler version, and include files were used to build it.
CCS, at various times, have claimed to be 'improving' their testing of new releases, but I've not seen any sign of it....
Learn not to think 'newer is better' with CCS...
Best Wishes |
|
|
Pekub
Joined: 20 Dec 2005
Posts: 16
|
|
| Posted: Wed Nov 23, 2011 8:34 am |
|
|
Thank`s PCM programmer.
The solution with "cast pointer" works right.
| Code: |
ptr = (int8 *)(pRx->Buff) +0; // Cast pointer to int8*
|
I think this relates to the CCS note in readme.txt:
Users that have old code with expresions of the form:
*(&data + i)
need to change them to:
*((int8 *)(&data) + i)
A compiler change was made to be ANSI compliant.
In any case, this is very troublesome change. I need "casting" all my program |
|
|
FvM
Joined: 27 Aug 2008
Posts: 2337
Location: Germany
|
|
| Posted: Wed Nov 23, 2011 10:09 am |
|
|
| Quote: | | I think this relates to the CCS note in readme.txt |
This is a quite different problem I think. It's about how pointers are increased, in bytes or size of the entity pointed to.
The problem revealed in your code is a "simple" bug in calculating pointer addresses. You can look to the generated assembly code, it's just rubbish, nothing about ANSI or previous CCS interpretation of pointers. If you abstract from the involved data types, three of the four pointer dereference cases are just an offset that can be calculated at compile time. CCS C however presents three different forms of calculation, more long-winded than necessary and two of them faulty.
Because of this "many different ways", there's a fair chance to make the compiler using a variant with correct result, as apparently achieved with the casting operation. Unfortunately, there's no systematics behind it. My personal rule of thumb is: When meeting a CCS bug, try simple instruction instead of complex one's, don't mind some extra lines of code. CCS C wants to be spoon-feeded. |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
Powered by phpBB © 2001, 2005 phpBB Group
|
FAQ 
Search
Register
Profile 
Log in to check your private messages 
Log in
