linux-kernel - Re: [RESEND][v2][PATCH] Fix a race between try_to_wake

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: <20160905074858.GP10153@twins.programming.kicks-ass.net>
Date:   Mon, 5 Sep 2016 01:48:58 -0600
From:   Peter Zijlstra <peterz@...radead.org>
To:     Benjamin Herrenschmidt <benh@...nel.crashing.org>
CC:     Balbir Singh <bsingharora@...il.com>,
        LKML <linux-kernel@...r.kernel.org>,
        Oleg Nesterov <oleg@...hat.com>,
        Nicholas Piggin <nicholas.piggin@...il.com>,
        Alexey Kardashevskiy <aik@...abs.ru>
Subject: Re: [RESEND][v2][PATCH] Fix a race between try_to_wake_up() and a woken up task

On Mon, Sep 05, 2016 at 05:14:19PM +1000, Benjamin Herrenschmidt wrote:
> On Mon, 2016-09-05 at 13:16 +1000, Balbir Singh wrote:
> 
> ï¿½.../...
> > 
> > Cc: Peter Zijlstra <peterz@...radead.org>
> > Cc: Nicholas Piggin <npiggin@...il.com>
> 
> Acked-by: Benjamin Herrenschmidt <benh@...nel.crashing.org>
> 
> > Signed-off-by: Balbir Singh <bsingharora@...il.com>
> > ---
> > ï¿½kernel/sched/core.c | 11 +++++++++++
> > ï¿½1 file changed, 11 insertions(+)
> > 
> > diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> > index 2a906f2..582c684 100644
> > --- a/kernel/sched/core.c
> > +++ b/kernel/sched/core.c
> > @@ -2016,6 +2016,17 @@ try_to_wake_up(struct task_struct *p, unsigned
> > int state, int wake_flags)
> > ï¿½	success = 1; /* we're going to change ->state */
> > ï¿½	cpu = task_cpu(p);
> > ï¿½
> > +	/*
> > +	ï¿½* Ensure we see on_rq and p_state consistently
> > +	ï¿½*
> > +	ï¿½* For example in __rwsem_down_write_failed(), we have
> > +	ï¿½*ï¿½ï¿½ï¿½ï¿½[S] ->on_rq = 1				[L] ->state
> > +	ï¿½*ï¿½ï¿½ï¿½ï¿½MB					ï¿½RMB
> > +	ï¿½*ï¿½ï¿½ï¿½ï¿½[S] ->state = TASK_UNINTERRUPTIBLE	[L] ->on_rq
> > +	ï¿½* In the absence of the RMB p->on_rq can be observed to be 0
> > +	ï¿½* and we end up spinning indefinitely in while (p->on_cpu)
> > +	ï¿½*/

So I did replace that comment with the one I proposed earlier. I checked
a fair number of architectures and many did not have an obvious barrier
in switch_to(). So that is not something we can rely on, nor do we need
to I think.

> > +	smp_rmb();
> > ï¿½	if (p->on_rq && ttwu_remote(p, wake_flags))
> > ï¿½		goto stat;
> > ï¿½