linux-kernel - Re: [PATCH 06/16] freezer: make exiting tasks properly unfreezable

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, 25 Aug 2011 17:25:33 +0200
From:	Oleg Nesterov <oleg@...hat.com>
To:	Matt Helsley <matthltc@...ibm.com>
Cc:	Tejun Heo <tj@...nel.org>, rjw@...k.pl, menage@...gle.com,
	linux-kernel@...r.kernel.org, arnd@...db.de
Subject: Re: [PATCH 06/16] freezer: make exiting tasks properly unfreezable

On 08/24, Matt Helsley wrote:
>
> On Fri, Aug 19, 2011 at 04:16:12PM +0200, Tejun Heo wrote:
> > There's no point in freezing an exiting task.  The current code
> > seemingly tries that by calling clear_freeze_flag() from exit_mm() but
> > it's racy as freeze might happen afterwards.
> >
> > This patch removes the racy clear_freeze_flag() makes do_exit() set
> > PF_NOFREEZE after PTRACE_EVENT_EXIT, after which freezing doesn't make
> > sense.
> >
> > Signed-off-by: Tejun Heo <tj@...nel.org>
> > ---
> >  kernel/exit.c          |    8 ++++++--
> >  kernel/power/process.c |    3 +--
> >  2 files changed, 7 insertions(+), 4 deletions(-)
> >
> > diff --git a/kernel/exit.c b/kernel/exit.c
> > index 2913b35..ac58259 100644
> > --- a/kernel/exit.c
> > +++ b/kernel/exit.c
> > @@ -679,8 +679,6 @@ static void exit_mm(struct task_struct * tsk)
> >  	tsk->mm = NULL;
> >  	up_read(&mm->mmap_sem);
> >  	enter_lazy_tlb(mm, current);
> > -	/* We don't want this task to be frozen prematurely */
> > -	clear_freeze_flag(tsk);
>
> This patch doesn't look quite right. Perhaps that's because I'm
> unclear why any of the freezer flags matter in the do_exit() path. How
> can the task enter the refrigerator in do_exit()?

it can't.

> If we can't enter the refrigerator then why fiddle with these flags
> here at all? That might explain why this race never introduced
> problems.

One problem is, try_to_freeze_tasks() shouldn't fail if we have a zombie.

> Couldn't TIF_FREEZE already be set?

Yes,

> Setting PF_NOFREEZE only ensures
> the flag won't get set "after" this point. To fix this I think we'd need
> something more like:
>
> 	current->flags |= PF_NOFREEZE;
> 	smp_wmb();
> 	clear_freeze_flag(tsk);

We don't really need this, we are not going to freeze.




However. I thought about this too. In the long term I think we do need
some cleanups to avoid the wrong TIF_SIGPENDING... but currently this
is off-topic.

Oleg.

--
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/