Re: glib 2.34 g_main_context_unref hangs

[Ken Bass <kbass kenbass com>]

On 01/18/2013 09:46 AM, Dan Winship wrote:
> On 01/17/2013 02:48 PM, Ken Bass wrote:
> > On 01/17/2013 01:58 PM, Dan Winship wrote:
> > > g_main_context_default() does not give you a ref on the default context,
> > > so your code here is destroying an object internal to glib. If you call
> > > this more than once (or, if you do *anything* involving the default
> > > context after this point), I'd definitely expect it to fail.
> > >
> > > Beyond that, it's hard to say what you might be doing wrong without
> > > seeing more code.
> > Odd that it worked all these years
> Yeah, probably you were just lucky before...
>
> > For some reason I was under the impression it gave me a reference,
> > particularly because
> > when I look at that default context, its ref_count is 1, not 0.
> Right, it's created with a ref_count of 1 when you first call
> g_main_context_default(), but every time you call it after that, it just
> returns the same one again without reffing it.
 I might not have been clear here. I meant the reference count was 
still 1 right before program exit.
> > My clean
> > up routine is the last thing executed before exit(), so
> > wouldn't this mean there is a memory leak?
> If you are only calling the cleanup routine once, then there shouldn't
> be any problem... is it possible you're freeing the GMainContext twice
> somehow?
Yes, I only call the cleanup routine once. But I call:
 g_main_loop_unref(main_loop);
 g_main_context_unref(context);

g_main_loop_unref() internally does unrefs the context (which should be 
the default context). So in essence the above is freeing it twice.
What I thought was odd was that after the g_main_loop_unref() call, the 
ref_count of the context was still 1.
Maybe I was confused and this was invalid memory left over from before 
the unref/free that I am seeing in gdb.


> Anyway, there is no memory "leak" (because glib still has a pointer to
> the context, and could still return it it to you if you called
> g_main_context_default() again). There is just memory that is left
> unfreed on exit, which is not a bug. (OTOH, if you are valgrinding and
> trying to get rid of all possible leaks, it is useful to be able to free
> some of that stuff, and I know that freeing the default GMainContext
> immediately before exit *does* work, because all of the libsoup test
> programs do that. So, there may be something else going wrong in your
> code, like trying to free it twice or something.)
I am pretty certain that my reason for freeing this many years ago was 
due to valgrind analysis and cleanup.

You say that you know freeing the default GMainContext works, does the 
libsoup test do both the
g_main_loop_unref(main_loop);
g_main_context_unref(default_context);

If so, that is no different that what I am doing, which would appear to 
be a double unref.

Thanks for your help. I ended up just checking if I was dealing with the 
default context and if so, do not
unref it since I do not own it. That works fine and does make sense.