If I place atexit( fn );
on the exit stack, it will get executed when the program exits: returns from main()
or via exit()
.
Can I remove it from the stack?
Why do I want to do this, you ask?
I was experimenting with a simple try-catch mechanism using atexit
, setjmp
and longjmp
. It would be just perfect if I could undo-atexit(fn);
- even if it would only work for the last registered function.
Edit:
Following monoceres' suggestion to make my own stack...
The stack only works with one exception catcher for now.
void (*_catchFn[10])() = {0,0,0,0,0,0,0,0,0,0};
void _catch(){
if ( _catchFn[0] != 0 ){
(_catchFn[0])();
}
}
void _addCatch( void (*fn)() ){
_catchFn[0]=fn;
}
void _remCatch( void (*fn)() ){
_catchFn[0]=0;
}
void test(){
jmp_buf env;
void catch(){ // we get here after an exit with a registered catch
longjmp(env,1); // return to the line marked except...
// that first will get the value 1
}
int first = setjmp( env); // ** return here **
fprintf( stderr , "test: After setjmp. first=%d\n" , first );
if( first == 0 ){ // try this code
_addCatch(catch); // register the catch function to 'catch' the exit
fprintf( stderr , "test: Before CHECK\n" );
// CHECK something and something bad happens and it exits
exit(1); // like this
fprintf( stderr , "test: After CHECK - THIS SHOULD NEVER BE SEEN AFTER AN EXCEPTION.\n" );
}else{
fprintf( stderr , "test: After longjmp return. first=%d\n" , first );
}
_remCatch( catch);
fprintf( stderr , "test: IT WORKED!\n");
exit(1); // exit again to see if we are safe
}
int main(){
atexit( _catch ); // register my global exception stack
test();
}
Why not build your own stack that you call from a single atexit() function? That way you could manipulate the stack all you want.