From: "Mauricio Fernández" <batsman.geo@...>
Date: 2003-09-04T06:51:33+09:00
Subject: Re: Ruby GC eating my VALUE on the stack

On Thu, Sep 04, 2003 at 03:43:21AM +0900, Sean O'Dell wrote:
> I thought I had read everything I needed about the Ruby GC, but 
> something just popped up that took me by surprise.
> 
> I am allocating a structure using malloc in QuiXML (in C), then wrapping 
> it in Data_Wrap_Struct, providing mark and free functions.
> 
> Data_Wrap_Struct returns a VALUE, which is stored in a variable on the 
> stack.
> 
> I never leave this function; I call other functions and pass the pointer 
> to the structure around, but the variable holding the VALUE stays on the 
> stack the entire time.
> 
> I am stress-testing QuiXML, so I am deliberately dogging memory to get 
> the GC to invoke naturally.
> 
> At some point, a rb_str_new() call causes the GC to kick in because 
> right then, the free callback for my structure gets called.
> 
> Now, from what I understand, since the VALUE associated with the 
> structure is still on the stack, the GC was supposed to see it and 
> automatically mark it so my structure won't get freed prematurely.
> 
> So why did it not get marked?  Why was free called?

Is there any difference if you declare the VALUE to be volatile?

-- 
 _           _                             
| |__   __ _| |_ ___ _ __ ___   __ _ _ __  
| '_ \ / _` | __/ __| '_ ` _ \ / _` | '_ \ 
| |_) | (_| | |_\__ \ | | | | | (_| | | | |
|_.__/ \__,_|\__|___/_| |_| |_|\__,_|_| |_|
	Running Debian GNU/Linux Sid (unstable)
batsman dot geo at yahoo dot com

The only other people who might benefit from Linux8086 would be owners
of PDP/11's and other roomsized computers from the same era.
	-- Alan Cox