Re: segmentation fault.

[Jonathan Vanasco <suppressed>]

This happened to me once (actually many times, but I figured out my  
approach on the first time), I could only figure out one way to  
handle it, and its been the one thing I hate about perl ever since  
(as I know there HAS to be a bettter way )

What I do:
	package MyApp::Debug;
	use constant DEBUG_FUNCTION_NAME => 1;
	sub print_function_name {
		# derive function name by looping caller, print to STDERR
	}
	sub print_caller_stack {
		# print the entire caller stack. useful for odd debugging, print to  
STDERR
	}

	package MyApp::JustAboutEverything;
	sub render_page {
		my 	( $self )= @_;
		MyApp::Debug::DEBUG_FUNCTION_NAME &&  
MyApp::Debug::print_function_name();
	}
	sub another_function {
		my 	( $self )= @_;
		MyApp::Debug::DEBUG_FUNCTION_NAME && MyApp::Debug::  
print_function_name ();
	}

its messy, because I need to put that damn DEBUG_FUNCTION_NAME in  
every sub I want to act like this.  if i were a better perl  
programmer, I'd have take the time to figure out a better way to  
overload how perl calls functions.

in any event, since DEBUG_FUNCTION_NAME is based a constant, the  
lines are nicely optimized away in production code ( assuming you  
turn that option off ).  i think it ends upon average to making my  
dev server run about 5% larger in memory use and about 5% slower in  
exeuction too.  but i just set DEBUG_FUNCTION_NAME to 0, and its gone.

my guess is that your error is causing an issue in some XS code or in  
perl/apache iteself - so you're not going to get any error caught.

in the past, i've gotten those segfaults from:
	reading a bad cookie ( libapreq2 .05, i think ) ( c error )
	handling an upload via apache request2 ( c error )
	having a bad tied session under Apache2::Session ( xs error )
	some random apache2 error.

i don't know if any of that helps.  good luck!
	

On May 23, 2007, at 3:49 PM, Tyler Bird wrote:

> Ok,
>
> I have a system I converted from cgi to mod_perl.
>
> We recently upgraded to mod_perl 2.0 and apache 2.2 on RHL5
>
> I am having a problem.  When I refresh a certain page 5 times or  
> about ( it's completely random )
>
> The page renders fine, but in my logs I see that an apache child  
> died because of a segmentation fault I believe
> here is the error or I guess notice message.
>
> [notice] child pid 25946 exit signal Segmentation fault (11) # ? do  
> I have to worry about this notice ? what we are building will be a  
> live system.
>
> My question is why is it doing this?  I have but try{} catch  
> clauses on the whole script and it has yielded me no errors
> that I have detected on this before like somewhere deep in the code  
> I am missing a require().
>
> Any ideas on how I could better trap a segmentation fault besides  
> using perl's try{} catch{} like  catching it's signal?