linux-kernel - Re: [PATCH v3 03/13] seccomp_filters: new mode with configurable syscall filters

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]

Date:	Thu, 02 Jun 2011 16:46:07 -0400
From:	Steven Rostedt <rostedt@...dmis.org>
To:	Will Drewry <wad@...omium.org>
Cc:	paulmck@...ux.vnet.ibm.com, linux-kernel@...r.kernel.org,
	kees.cook@...onical.com, torvalds@...ux-foundation.org,
	tglx@...utronix.de, mingo@...e.hu, jmorris@...ei.org,
	Peter Zijlstra <peterz@...radead.org>,
	Frederic Weisbecker <fweisbec@...il.com>,
	linux-security-module@...r.kernel.org
Subject: Re: [PATCH v3 03/13] seccomp_filters: new mode with configurable
 syscall filters

On Thu, 2011-06-02 at 15:28 -0500, Will Drewry wrote:

[ Snipped 860 lines of non relevant text ]

Seriously guys, Please trim your replies. These last few messages were
ridicules. I spent more than 30 seconds searching for what the email was
about. That's too much wasted time.

-- Steve


> >> Ah, I spaced on rcu_dereference().  The goal was to make the
> >> assignment and replacement of the seccomp_filters pointer
> >> RCU-protected (in seccomp_state) so there's no concern over it being
> >> replaced partial on platforms where pointer assignments are non-atomic
> >> - such as via /proc/<pid>/seccomp_filters access or a call via the
> >> exported symbols.  Object lifetime is managed by reference counting so
> >> that I don't have to worry about extending the RCU read-side critical
> >> section by much or deal with pre-allocations.
> >>
> >> I'll add rcu_dereference() to all the get_seccomp_filters() uses where
> >> it makes sense, so that it is called safely.  Just to make sure, does
> >> it make sense to continue to rcu protect the specific pointer?
> >
> > It might.  The usual other options is to use a lock outside of the element
> > containing the reference count to protect reference-count manipulation.
> > If there is some convenient lock, especially if it is already held where
> > needed, then locking is more straightforward.  Otherwise, RCU is usually
> > a reasonable option.
> 
> I was concerned about the overhead a lock would have at each system
> call entry, but I didn't benchmark it to see.  I'll add the
> rcu_dereference right away, then look into seeing whether there's a
> cleaner approach.  I was trying to be overly protective of mutating
> any data internal to the filters through complete replacement on any
> change.  I'll take a step back and see if
> 


--
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/