Date: Wed, 30 Jul 2008 10:09:38 -0400
From: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
To: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>, linux-kernel@vger.kernel.org,
       akpm@linux-foundation.org
Subject: Re: [PATCH] Module : call synchronize_sched() between module
	exit() and free.
Message-ID: <20080730140938.GC18210@Krystal>
References: <20080717155724.897537670@polymtl.ca> <20080729211543.GB17097@redhat.com> <20080729224115.GA5209@Krystal> <200807301140.59745.rusty@rustcorp.com.au> <20080730022751.GA15866@Krystal> <20080730114035.GD27711@redhat.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
In-Reply-To: <20080730114035.GD27711@redhat.com>
User-Agent: Mutt/1.5.16 (2007-06-11)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 2084
Lines: 66

* Frank Ch. Eigler (fche@redhat.com) wrote:
> Hi -
> 
> On Tue, Jul 29, 2008 at 10:27:51PM -0400, Mathieu Desnoyers wrote:
> > [...]
> > Actually, it's not placing a marker/tracepoint in a module which causes
> > a problem, this is a simple function call after all, and correctly dealt
> > with by current module.c code.
> > [...]
> 
> Just to spell it out, it is this scenario I'd like to see documented:
> 
> module-foo.c:
>    foo() { ... trace_mark (foo, "..."); ... }
> 
> module-bar.c:
>    setup() { ... marker_probe_register ("foo" , ..., &foo_handler ); }
>    teardown() { ... marker_probe_unregister ("foo" , ..., &foo_handler ); }
>    foo_handler() { }
> 
> 1)  module-foo loads
> 2)  module-bar loads
> 3)  module-bar.c:setup()
> 4)  module-foo unloads
> 
> What happens here?  Certainly no more calls to foo_handler, but is
> that all?

Yep, that's it. However a hash table still keeps track of the "foo"
handler so that it's reconnected whenever module-foo is reloaded.

>(Would it not be desirable for an active marker to cause
> module-foo's refcount to increase, so as to prevent unloading at this
> time?)

No, because I want to be able to unload the marked module and I don't
want the fact that a probe is connected to it to change that.

> 
> 5)  module-bar.c:teardown()
> 
> Can this teardown code succeed fully even if module-foo is already
> dead and gone?
>   

Yes.

Actually there is a detail missing here. Your teardown should be :
 teardown() { ... marker_probe_unregister ("foo" , ..., &foo_handler );
                  synchronize_sched(); /* Before returning from exit */ }

This makes sure that every live marker call are finished and that it is
safe to unload module-bar (the probe).

Mathieu
> 
> - FChE

-- 
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/