View previous topic :: View next topic |
Author |
Message |
Gabriel
Joined: 03 Aug 2009 Posts: 1067 Location: Panama
|
Phantom Function calls |
Posted: Mon Apr 01, 2019 12:32 pm |
|
|
Hi, I have a Gigantic project, many thousands of lines of code, and I'm getting into a strange situation.
I'm getting phantom function calls.
For example, a particular section of code runs and out of nowhere an unrelated function gets called. So far its always the same function. I can repeat the error by making the device run certain code paths with obviously do not involve calling the phantom function.
Fortunately the phantom function call is pretty benign, the only "real" consequence is a slight delay, since this function needs to execute.
But I need to solve this.
I have seen weird errors like this due to too many int1 variables, but in this particular case i believe its due to a particular long function in my main .C file. I believe this particular function might be stepping over some memory boundary and the compiler is having a fit. How can i check?
This gigantic function is a huge switch statement with is responsible for "typing" text messages (different kinds of alerts, errors, etc), but i have over 20 cases with multiple printf statements each case.
So, any suggestions on what could be causing phantom calls or how can i reduce a switch with many cases?
Thanks
G. _________________ CCS PCM 5.078 & CCS PCH 5.093 |
|
![](templates/subSilver/images/spacer.gif) |
Ttelmah
Joined: 11 Mar 2010 Posts: 19369
|
|
Posted: Mon Apr 01, 2019 12:49 pm |
|
|
How many cases?.
Honestly 5.013, is a very early V5 compiler before several bugs
were fixed.
If you have no 'default' case, and the cases are sequential on a
PIC16 or 18, the switches should code as a jump table. With 5.013
it is possible you are experiencing a page switch problem from the
jump table. |
|
![](templates/subSilver/images/spacer.gif) |
PCM programmer
Joined: 06 Sep 2003 Posts: 21708
|
|
Posted: Mon Apr 01, 2019 1:03 pm |
|
|
Download the CCS demo. Presumably it's the latest version, or near to it.
See if you still get the problem. If not, then upgrade. Or buy the 1-year
maintenance if they allow you to. |
|
![](templates/subSilver/images/spacer.gif) |
Gabriel
Joined: 03 Aug 2009 Posts: 1067 Location: Panama
|
|
Posted: Mon Apr 01, 2019 1:08 pm |
|
|
17 actually coded (its a work in progress) + Default case.
Cases in this switch are controlled by a single int with values 1 - 35
cases 1-14 are coded, then a jump to cases 20-21 + Default.
to be clear i have not defined cases 15-20.
I have sitting a folder 5.078 which i have not installed because i said i was not going to update until i finished this project because i didnt want to deal with any possible bugs from the compiler change... also because i forgot how to update.
I really appreciate your help. _________________ CCS PCM 5.078 & CCS PCH 5.093 |
|
![](templates/subSilver/images/spacer.gif) |
dluu13
Joined: 28 Sep 2018 Posts: 395 Location: Toronto, ON
|
|
Posted: Mon Apr 01, 2019 1:22 pm |
|
|
ah... I'm on 5.082, and I have had some of these "phantom" bits of code that get executed. I was confused, but I never looked into it though...
I just changed them to int8 0's and 1's and then stopped having issues.
I know this is off topic, but why are the int1's misbehaving? |
|
![](templates/subSilver/images/spacer.gif) |
temtronic
Joined: 01 Jul 2010 Posts: 9174 Location: Greensville,Ontario
|
|
Posted: Mon Apr 01, 2019 1:30 pm |
|
|
Quick, easy test...
I'd define all of those undefined cases. Just some 'nop' do nothing code but fill the table... they become 'place holders' that organise the 'code' And... presumably you'll need them in the near future...
IF the phantom menace goes away, you've got your answer. If not, well, you've eliminated it as being the cause.
As for int1 misbehaving.. I think I read somewhere, long, long time ago it's best to put 8 int1s into an int8. If you need 7 int1, add a dummy int1 to fill a byte.
Jay |
|
![](templates/subSilver/images/spacer.gif) |
Gabriel
Joined: 03 Aug 2009 Posts: 1067 Location: Panama
|
|
Posted: Mon Apr 01, 2019 1:32 pm |
|
|
Quote: | I think I read somewhere, long, long time ago it's best to put 8 int1s into an int8. If you need 7 int1, add a dummy int1 to fill a byte. |
This worked for me when i had this issue.
I will update to 5.078 and see what happens. _________________ CCS PCM 5.078 & CCS PCH 5.093 |
|
![](templates/subSilver/images/spacer.gif) |
PrinceNai
Joined: 31 Oct 2016 Posts: 463 Location: Montenegro
|
|
Posted: Mon Apr 01, 2019 1:48 pm |
|
|
I never liked those int1-s. Not because I had any problems with them, but for me it is way easier to check for a value of a flag with debugger if it is declared as int8. I'd only use them if I'd need every byte of RAM, which for my projects isn't an issue at all. |
|
![](templates/subSilver/images/spacer.gif) |
Gabriel
Joined: 03 Aug 2009 Posts: 1067 Location: Panama
|
|
Posted: Mon Apr 01, 2019 2:16 pm |
|
|
So I updated to 5.078.
I now get an error saying im missing an "#ENDIF" at the bottom of my main C file... im not.
ive added the endif just to see what happens and i get an error over a commented out block of code.
sigh... this is why i didnt want to update. _________________ CCS PCM 5.078 & CCS PCH 5.093 |
|
![](templates/subSilver/images/spacer.gif) |
Gabriel
Joined: 03 Aug 2009 Posts: 1067 Location: Panama
|
|
Posted: Mon Apr 01, 2019 2:35 pm |
|
|
Reverted back to 5.013 and it compiles just fine.
I did find that 5.078 found an error like this:
Code: | fprintf(lcd_putc"\r\n"); |
Note missing comma after stream - These have been corrected.
What now? _________________ CCS PCM 5.078 & CCS PCH 5.093 |
|
![](templates/subSilver/images/spacer.gif) |
PCM programmer
Joined: 06 Sep 2003 Posts: 21708
|
|
Posted: Mon Apr 01, 2019 2:42 pm |
|
|
I would have corrected the #endif problem. |
|
![](templates/subSilver/images/spacer.gif) |
Gabriel
Joined: 03 Aug 2009 Posts: 1067 Location: Panama
|
|
Posted: Mon Apr 01, 2019 2:46 pm |
|
|
PCM_P: that was my first path to resolving this... I have checked all included files for a missing #endif and i cant find it.
Its asking me for an #endif on a blank line after the last line of code.
its also giving me a warning on string.h which i find strange.
Quote: | Executing: "C:\Program Files (x86)\PICC - 5.078\Ccsc.exe" +FH "Heimdall - Test Program.c" +DF +LN +T +A +M +Z +Y=9 +EA #__18F87J50=TRUE
Compiling C:\Users\Foo-Bar\Documents\1. PIC Developement\CCS Compiler\CONASA\CONASA - HEIMDALL\Heimdall - Test Program on 01-abr-19 at 15:40
>>> Warning 236 "C:\Program Files (x86)\PICC - PCH - 5.013\drivers\string.h" Line 54(9,16): Defined identifier matches a variable ID
>>> Warning 230 "C:\Users\Foo-Bar\Documents\1. PIC Developement\CCS Compiler\CONASA\CONASA - HEIMDALL\H-GSM-General Driver-3G.c" Line 660(1,1): String truncated
>>> Warning 203 "C:\Users\Foo-Bar\Documents\1. PIC Developement\CCS Compiler\CONASA\CONASA - HEIMDALL\H-GSM-Amenities-3G.c" Line 74(1,1): Condition always TRUE
*** Error 14 "C:\Users\Foo-Bar\Documents\1. PIC Developement\CCS Compiler\CONASA\CONASA - HEIMDALL\Heimdall - Test Program.c" Line 1624(9,10): Missing #ENDIF
1 Errors, 3 Warnings.
Build Failed.
Halting build on first failure as requested.
BUILD FAILED: Mon Apr 01 15:40:37 2019 |
_________________ CCS PCM 5.078 & CCS PCH 5.093 |
|
![](templates/subSilver/images/spacer.gif) |
PCM programmer
Joined: 06 Sep 2003 Posts: 21708
|
|
Posted: Mon Apr 01, 2019 2:56 pm |
|
|
I uninstalled vs. 5.083, and installed vs. 5.078. I then compiled the
following program. It didn't give any errors or warnings.
I'm wondering if you have 5.013 CCS driver files and #include files
copied to your local project folder. Or somehow you are using a mix
of old files and new ones from 5.078.
Test program:
Code: | #include <18F87J50.h>
#fuses NOWDT
#use delay(crystal=20M)
#include <string.h>
//==========================
void main()
{
while(TRUE);
}
|
|
|
![](templates/subSilver/images/spacer.gif) |
Gabriel
Joined: 03 Aug 2009 Posts: 1067 Location: Panama
|
|
Posted: Mon Apr 01, 2019 3:14 pm |
|
|
Quote: | I'm wondering if you have 5.013 CCS driver files and #include files
copied to your local project folder. Or somehow you are using a mix
of old files and new ones from 5.078. |
This might explain the string.h warning.
But the #endif is killing me.. ive re-checked all my code.. im not missing one.
The 2 most extreme cases of #IF type of statements i use are as follows:
Code: |
#ifdef SUPER_CLICK
#define ESP_RESET PIN_A1 //Clasic ESP (cheap one)
#else
#ifdef LEGACY
#define ESP_RESET PIN_A1 //Original MikroE ESP
#else
#define ESP_RESET PIN_D1 //New MikroE ESP
#endif
#endif |
Code: |
#ifdef LEGACY
if(Find_in_Buffer("OK",20))
#else
if(Find_in_Buffer("GOT IP",20))
#endif
{
fprintf(lcd_putc,"WIFI - Connected to Network\r\n");
WIFI_RESET_FLAG=FALSE;
return(1);
}
else return(0); |
This is part of my ESP-8266 driver, to handle older versions of the hardware.
This has worked for the last 4 years...across several projects that share this driver... although it works... ive always felt its ugly code (the IF statement selection)...
I cant find a missing #ENDIF... _________________ CCS PCM 5.078 & CCS PCH 5.093 |
|
![](templates/subSilver/images/spacer.gif) |
Gabriel
Joined: 03 Aug 2009 Posts: 1067 Location: Panama
|
|
Posted: Mon Apr 01, 2019 3:34 pm |
|
|
Wait a minute:
Quote: | >>> Warning 236 "C:\Program Files (x86)\PICC - PCH - 5.013\drivers\string.h" Line 54(9,16): Defined identifier matches a variable ID |
how do i correct this so that it pulls it from the the newly installed 5.078 folder? _________________ CCS PCM 5.078 & CCS PCH 5.093 |
|
![](templates/subSilver/images/spacer.gif) |
|