Subject: Re: [PATCH] block2mtd lockdep_init_map warning
From: Peter Zijlstra <a.p.zijlstra@chello.nl>
To: Rusty Russell <rusty@rustcorp.com.au>
Cc: Peter Zijlstra <peterz@infradead.org>, Erez Zadok <ezk@cs.sunysb.edu>,
       =?ISO-8859-1?Q?J=F6rn?= Engel <joern@logfs.org>, dwmw2@infradead.org,
       linux-mtd@lists.infradead.org, mingo@redhat.com,
       linux-kernel@vger.kernel.org
In-Reply-To: <200801081147.00701.rusty@rustcorp.com.au>
References: <200801061911.m06JBldW020012@agora.fsl.cs.sunysb.edu>
	 <1199700326.7143.10.camel@twins> <200801081147.00701.rusty@rustcorp.com.au>
Content-Type: text/plain
Date: Wed, 16 Jan 2008 09:16:38 +0100
Message-Id: <1200471398.9535.24.camel@taijtu>
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 3078
Lines: 83


On Tue, 2008-01-08 at 11:47 +1100, Rusty Russell wrote:
> On Monday 07 January 2008 21:05:26 Peter Zijlstra wrote:
> > On Sun, 2008-01-06 at 14:11 -0500, Erez Zadok wrote:
> > > > Ingo, Peter, does either of you actually care about this problem?  In
> > > > the last round when I debugged this problem there was a notable lack of
> > > > reaction from either of you.
> > >
> > > The problem appears to be an interaction of two components--module
> > > loading and lockdep--that's perhaps why it wasn't given enough attention.
> >
> > Would something like this work for people?
> 
> Hi Peter,
> 
>     There's nothing wrong with this patch, but I think it papers over a more
> general problem: we enter the module (to parse args) while it's not in the
> module list.  This also means we won't get a nice oops if it crashes.
> 
>     This is untested, but does it solve it for you?

I think it should, Erez care to give it a spin?

> diff -r 68fd1b22db89 kernel/module.c
> --- a/kernel/module.c	Mon Jan 07 18:59:50 2008 +1100
> +++ b/kernel/module.c	Tue Jan 08 11:46:11 2008 +1100
> @@ -2043,6 +2043,11 @@ static struct module *load_module(void _
>  		printk(KERN_WARNING "%s: Ignoring obsolete parameters\n",
>  		       mod->name);
>  
> +	/* Now sew it into the lists so we can get lockdep and oops
> +         * info during argument parsing.  Noone should access us, since
> +         * strong_try_module_get() will fail. */
> +	stop_machine_run(__link_module, mod, NR_CPUS);
> +
>  	/* Size of section 0 is 0, so this works well if no params */
>  	err = parse_args(mod->name, mod->args,
>  			 (struct kernel_param *)
> @@ -2051,7 +2056,7 @@ static struct module *load_module(void _
>  			 / sizeof(struct kernel_param),
>  			 NULL);
>  	if (err < 0)
> -		goto arch_cleanup;
> +		goto unlink;
>  
>  	err = mod_sysfs_setup(mod,
>  			      (struct kernel_param *)
> @@ -2059,7 +2064,7 @@ static struct module *load_module(void _
>  			      sechdrs[setupindex].sh_size
>  			      / sizeof(struct kernel_param));
>  	if (err < 0)
> -		goto arch_cleanup;
> +		goto unlink;
>  	add_sect_attrs(mod, hdr->e_shnum, secstrings, sechdrs);
>  	add_notes_attrs(mod, hdr->e_shnum, secstrings, sechdrs);
>  
> @@ -2074,7 +2079,8 @@ static struct module *load_module(void _
>  	/* Done! */
>  	return mod;
>  
> - arch_cleanup:
> + unlink:
> +	stop_machine_run(__unlink_module, mod, NR_CPUS);
>  	module_arch_cleanup(mod);
>   cleanup:
>  	module_unload_free(mod);
> @@ -2130,10 +2136,6 @@ sys_init_module(void __user *umod,
>  		mutex_unlock(&module_mutex);
>  		return PTR_ERR(mod);
>  	}
> -
> -	/* Now sew it into the lists.  They won't access us, since
> -           strong_try_module_get() will fail. */
> -	stop_machine_run(__link_module, mod, NR_CPUS);
>  
>  	/* Drop lock so they can recurse */
>  	mutex_unlock(&module_mutex);

--
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/