My sketch (pretty big - 100kb complied on Mega) uses number of global, local and static variables.
Below is a fraction of the globals :
unsigned long timePB; // holds the time the address push button is pressed
char inputChar[10]; // Manual Input from Serial port
byte count=1; // used for sending parts of received packets to the Log function
byte timerExecuted[21]; // Stores the state of each of the timers
int input; // Temporary variables to hold user input from serial/telnet
int input2;
char filename[16]; // stores filename for log
byte sendPending=0; // Indicates if SendFeedback function has postponed bus transmission because of bus congestion
unsigned long packetsDropped=0; // Number of packets dropped due to bus congestion
byte SkipTelnetInput=0; // When raised (1) Telnet input is skipped becasue a command is in process
byte SDCARDValid=0; // SD card status
byte warned=0; // Telnet warning flag
byte messages=0; // register to hold active messages
byte statusRun =0; // System running status register
byte statusGen =0;
byte statusMech=0; // Status register for Mechanical systems
byte switchDevices [10]; // Holds switching devices
byte devInterNo=0; // Device index to interrogate from array above
unsigned long overflowpackets=0; // register to hold number of packets with additional AA start characters
unsigned long totalSent=0; // Number of command packets sent
unsigned long totalRetries[4] ={0,0,0,0}; // Number of packets that had to be resent
unsigned long ledflash=1;
Now, totalRetries[0] gets assigned the value of 6315870 with no obvious reason.
Please note that the value above is always the same and doesn't seem to change over the sketch execution. I have removed all calculations that assign values to that variable and its contents still the same.
I therefore suspect a memory leak. Possibly some array subscript going out of range.
Is there a way to pinpoint the course of this problem ?
How does the compiler assign memory locations to the arrays?
For example which variable would be held at the memory location just before totalRetries[0] on the memory map?
totalRetries[sendingTries[index])++;This doesn't compile anyway due to the 2x '[' and 1 ']' & 1 ')' so to figure out if it actually does what it should do we'd need to see the exact copy. Point is though if 'index' is ever bigger then '3' what will happen ?
static byte sendingTries[5]={1,1,1,1,1};
int resendTimeout=400;
for (byte index =0; index<5; index++){
if ( millis()- replyTimeout[index] > resendTimeout * sendingTries[index] && replyTimeout[index] > 0 && sendingTries[index] <4) {
resendCommand(index);
totalRetries[sendingTries[index]]++;
sendingTries[index]++;
return;
}
}
Basicaly what it does is, check all 4 available slots for timeouts. If any of them have timed out, it resends that packet, and increments counters.
Now the interesting part (I just found that out) is that this 'memory leak' happens even if the above code doesn't get executed at all!!
If I comment out " totalRetries[byte(sendingTries[index])]++; ", then everything is fine.
If the line is left there , even if not executed (because the If clause is not true), then totalRetries[0] somehow gets the value 6315870 straight away.
If I comment out " totalRetries[byte(sendingTries[index])]++; ", then everything is fine.
If the line is left there , even if not executed (because the If clause is not true), then totalRetries[0] somehow gets the value 6315870 straight away.
it must be the only reference to totalRetries within that function, it is very possible that somewhere else (outside of the snippet that you posted) within that function lies the key to the issue. Btw millis()- replyTimeout[index] > resendTimeout * sendingTries[index] && replyTimeout[index] > 0 && sendingTries[index] <4)for efficiencies sake i would do the test in the order that makes them go false the quickest, and for overviews sake i would 'brace' every condition between the '&&'
Watcher:
If the line is left there , even if not executed (because the If clause is not true), then totalRetries[0] somehow gets the value 6315870 straight away.
What are the four things declared before totalRetries? static data is included. (You may have to generate a MAP file or use AVR-NM to answer the question.)
it must be the only reference to totalRetries within that function, it is very possible that somewhere else (outside of the snippet that you posted) within that function lies the key to the issue. Btw
Thought of that too. But no, I did a search
for efficiencies sake i would do the test in the order that makes them go false the quickest
,
Forced them all false and tried it...
and for overviews sake i would 'brace' every condition between the '&&'
Used to to that but I now find the additional 'brace' just extra clutter and dont do it any more...(personal taste i guess)
For some (incorrect) definition of "whole", maybe.
Point taken PaulS.
However I tried to put the minimum code necessary for people to go through within reasonable amount of time without having to go though some 100kb of compiled code spanning over 19 file (IDE) tabs.
Watcher:
If the line is left there , even if not executed (because the If clause is not true), then totalRetries[0] somehow gets the value 6315870 straight away.
Ah. Then you need to divide-and-conquer to identify the offending section.
Riddle your code with calls to (something like) this...
