Locating 2 parameters at the same address

To use the keyword "volatile" was my first thought, and it worked for a while in my program. In this simplistic program it works as well. As I continued working with my code the problem got "unfixed".

I think I've demonstrated the problem here and I'm not confident that this solution with "volatile" is at all guaranteed. I'd guess that we a bit afield of the ANSI standard it it's up to how Keil designed their optimization.

I'll probably send this over to Keil technical support.

To use the keyword "volatile" was my first thought, and it worked for a while in my program. In this simplistic program it works as well. As I continued working with my code the problem got "unfixed".

I think I've demonstrated the problem here and I'm not confident that this solution with "volatile" is at all guaranteed. I'd guess that we a bit afield of the ANSI standard it it's up to how Keil designed their optimization.

I'll probably send this over to Keil technical support.

Another thing is to disable passing parameters in registers, and make sure that the functions do something:

#pragma NOREGPARMS
#include "c8051f9xx.h"

typedef unsigned char U8;
typedef unsigned int U16;

U8 Sub1(  U8 SubVar1 )          // No 'volatile'
{
        SubVar1 = SubVar1 + 1;
        return SubVar1;
}

U8 Sub2(  U8 SubVar2 )          // No 'volatile'
{
        SubVar2 = SubVar2 + 1;
        return SubVar2;
}


void main(void)
{
        while(1)
        {
                Sub1( 13 );
        }
}


static void ADC_ISR(void) interrupt 10
{
        U16 u16Result = 12;
        Sub2( u16Result );
        ADC0CN &= ~0x20;
}

Which then gives:

             ; FUNCTION Sub1 (BEGIN)
                                           ; SOURCE LINE # 7
                                           ; SOURCE LINE # 8
                                           ; SOURCE LINE # 9
0000 0500        R     INC     SubVar1
                                           ; SOURCE LINE # 10
0002 AF00        R     MOV     R7,SubVar1
                                           ; SOURCE LINE # 11
0004         ?C0001:
0004 22                RET
             ; FUNCTION Sub1 (END)

             ; FUNCTION Sub2 (BEGIN)
                                           ; SOURCE LINE # 13
                                           ; SOURCE LINE # 14
                                           ; SOURCE LINE # 15
0000 0500        R     INC     SubVar2
                                           ; SOURCE LINE # 16
0002 AF00        R     MOV     R7,SubVar2
                                           ; SOURCE LINE # 17
0004         ?C0002:
0004 22                RET
             ; FUNCTION Sub2 (END)

and

-------         MODULE        OVERLAY
  C:0000H         SYMBOL        _ICE_DUMMY_
  D:00E8H         PUBLIC        ADC0CN
  D:00A8H         PUBLIC        IE
  C:0003H         SYMBOL        ADC_ISR
  D:00B8H         PUBLIC        IP
  C:0024H         PUBLIC        main
  D:0088H         PUBLIC        TCON
  C:002CH         PUBLIC        Sub1
  C:0031H         PUBLIC        Sub2
  D:0098H         PUBLIC        SCON0
  D:00D0H         PUBLIC        PSW
  -------         PROC          SUB1
  D:0008H         SYMBOL        SubVar1
  C:002CH         LINE#         7
  C:002CH         LINE#         8
  C:002CH         LINE#         9
  C:002EH         LINE#         10
  C:0030H         LINE#         11
  -------         ENDPROC       SUB1
  -------         PROC          SUB2
  D:0009H         SYMBOL        SubVar2
  C:0031H         LINE#         13
  C:0031H         LINE#         14
  C:0031H         LINE#         15
  C:0033H         LINE#         16
  C:0035H         LINE#         17
  -------         ENDPROC       SUB2

"I think I've demonstrated the problem here"

I am not as confident.

I have tried to replicate your code, using a timer isr instead of your adc isr. Both variables are allocated at the same address.

and I never observed them having the same value.

so unless shown otherwise, I am going to agree with others that you are trying to solve a problem that doesn't exist.

No, I don't think that you have demonstrated it here.

That may or may not mean that there is no problem - it might just be because the example presented here is over-simplified so that the problem doesn't arise.

Probably worth a discussion with Keil support - likely to be easier to share a more substantial example with them...

I've tried to make my example more "complicated" but that darn optimizer is too smart. On the other hand my actual code is more than one swallow. I'll see if I can work up an example that actually fails.

Make sure your functions make use of the parameter, and have them write the parameter value to a volatile variable. Volatile variables are normally a good way to get the optimizer to stop removing code - you want to convince the compiler that there is an important side effect that must not be optimized away.

I've tried to make my example more "complicated" but that darn optimizer is too smart

try optimize = 2. this is basically variable overlay and nothing else.

to the homewrecker and the sardine: please not the word basically

Erik

"please not the word basically"

is it too much to ask that you write complete and comprehensible sentences, erik?

"I'll see if I can work up an example that actually fails."

I have one that follows your basic structure but uses a tmr isr to trigger one of the subroutines.

as andrew pointed out, you need to make sure that both subroutines actually operate on those two variables.

Folks, I think you are gallantly galloped way, way off the beaten path if the original poster is posting summary's like "A helpful post"

I love a good name calling and the pedantic arguments session, but maybe it would be better to create a new thread next time we want to do this? I do not think it was benefiting the OP.

Scratch that, I bet we can get this done without the name calling, especially the dreaded "s" word (sardine).