linux-kernel - Re: [RFC PATCH -tip ] [BUGFIX] x86: Remove preempt disabling from kprobes

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:	Fri, 01 Jul 2011 09:43:44 -0400
From:	Steven Rostedt <rostedt@...dmis.org>
To:	Masami Hiramatsu <masami.hiramatsu.pt@...achi.com>
Cc:	linux-kernel@...r.kernel.org,
	Peter Zijlstra <a.p.zijlstra@...llo.nl>,
	Frederic Weisbecker <fweisbec@...il.com>,
	Thomas Gleixner <tglx@...utronix.de>,
	Ingo Molnar <mingo@...e.hu>,
	Andrew Morton <akpm@...ux-foundation.org>,
	yrl.pp-manager.tt@...achi.com
Subject: Re: [RFC PATCH -tip ] [BUGFIX] x86: Remove preempt disabling from
 kprobes

On Fri, 2011-07-01 at 22:14 +0900, Masami Hiramatsu wrote:
> Steven Rostedt reported that putting kprobe on the instruction which
> loads preempt_count causes the wrong result, as below.
> 
> Kprobes requires preemption to be disabled as it single steps the code
> it replaced with a breakpoint. But because the code that is single
> stepped could be reading the preempt count, the kprobe disabling of the
> preempt count can cause the wrong value to end up as a result. Here's an
> example:
> 
> If we add a kprobe on a inc_preempt_count() call:
> 
> 	[ preempt_count = 0 ]
> 
> 	ld preempt_count, %eax	<<--- trap
> 
> 		<trap>
> 		preempt_disable();
> 		[ preempt_count = 1]
> 		setup_singlestep();
> 		<trap return>
> 
> 	[ preempt_count = 1 ]
> 
> 	ld preempt_count, %eax
> 
> 	[ %eax = 1 ]
> 
> 		<trap>
> 		post_kprobe_handler()
> 			preempt_enable_no_resched();
> 			[ preempt_count = 0 ]
> 		<trap return>
> 
> 	[ %eax = 1 ]
> 
> 	add %eax,1
> 
> 	[ %eax = 2 ]
> 
> 	st %eax, preempt_count
> 
> 	[ preempt_count = 2 ]
> 
> 
> We just caused preempt count to increment twice when it should have only
> incremented once, and this screws everything else up.
> 
> To solve this, I've removed preempt disabling code from kprobes,
> since the breakpoint exception and kprobes single step routine
> disables interrupts, it doesn't need to disable preemption while
> single-stepping anymore.
> 
> This patch is for -tip tree, and it can be applied to linus tree too.

I applied it to v3.0-rc5. And not surprisingly it works. But the
question I have is, when we return from the trap, and NEED_RECHED is
set, will it schedule? My test placed the probe within the scheduler
where preemption is already disabled. Let me do this in places that has
preemption and interrupts enabled.

I'll also try a kernel mod that adds a probe handler that does a
udelay() loop, forcing the timer interrupt to be set on return of the
trap, and see what happens there.

-- Steve


--
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/