linux-kernel - Re: [PATCH 6/7] Ksplice: Export symbols needed for Ksplice

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <alpine.DEB.2.00.0902080129020.11936@vinegar-pot.mit.edu>
Date:	Thu, 12 Feb 2009 20:46:40 -0500 (EST)
From:	Tim Abbott <tabbott@....EDU>
To:	Rusty Russell <rusty@...tcorp.com.au>
cc:	Jeff Arnold <jbarnold@....edu>,
	Andrew Morton <akpm@...ux-foundation.org>,
	linux-kernel@...r.kernel.org,
	Denys Vlasenko <vda.linux@...glemail.com>,
	Anders Kaseorg <andersk@....edu>,
	Waseem Daher <wdaher@....edu>,
	Nikanth Karthikesan <knikanth@...e.de>
Subject: Re: [PATCH 6/7] Ksplice: Export symbols needed for Ksplice

On Wed, 4 Feb 2009, Rusty Russell wrote:

> On Saturday 06 December 2008 10:33:59 Jeff Arnold wrote:
> > From: Tim Abbott <tabbott@....edu>
> > 
> > Ksplice uses several functions from module.c in order to resolve
> > symbols and implement dependency handling.  Calling these functions
> > requires holding module_mutex, so it is exported.
> 
> I have fought every attempt to export that.  It was long a cause of misery
> and pain; people saw the lock and so started grabbing it.
> 
> Any chance we can find another way?

It is definitely possible, but it wouldn't be pretty.  There are a few 
things that we use module_mutex for:

(1) Preventing races involving modules being unloaded during run-pre 
matching.  Functions such as kallsyms_on_each_symbol, find_symbol, 
__module_text_address, __module_address, and find_module that are used to 
obtain a module or a symbol can only be used within preempt_disable or 
while holding module_mutex.  Further, the values that they return can only 
be used in the same such context, unless one module_gets the relevant 
module in that context.

So, we'd change/add wrappers to a bunch of these functions to do their own 
preempt_disable or grabbing of module_mutex, and to module_get the 
relevant module before they return.  We'd also need to change the Ksplice 
code to module_put whenever we stop referencing a symbol.  Since a lot of 
the Ksplice kernel code is related to resolving symbols, I suspect this 
latter part would be somewhat messy.

(2) Calling use_module.  use_module needs to be called while holding 
module_mutex because it changes the modules_which_use_me lists.

We could change this by creating a modified version of use_module that 
grabs module_mutex itself; perhaps renaming the current use_module to 
__use_module.

(3) Preventing races involving two different Ksplice updates being applied 
at the same time.

This one is easy; we could just create a new ksplice_mutex.

	-Tim Abbott
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/