linux-kernel - Re: [RFC][PATCH] HWPOISON: remove the unsafe __set_page

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:	Mon, 28 Sep 2009 06:29:58 +0200
From:	Nick Piggin <npiggin@...e.de>
To:	Andi Kleen <andi@...stfloor.org>
Cc:	Hugh Dickins <hugh.dickins@...cali.co.uk>,
	Wu Fengguang <fengguang.wu@...el.com>,
	Andrew Morton <akpm@...ux-foundation.org>, linux-mm@...ck.org,
	LKML <linux-kernel@...r.kernel.org>
Subject: Re: [RFC][PATCH] HWPOISON: remove the unsafe __set_page_locked()

On Mon, Sep 28, 2009 at 06:11:08AM +0200, Andi Kleen wrote:
> On Mon, Sep 28, 2009 at 04:57:41AM +0200, Nick Piggin wrote:
> > On Mon, Sep 28, 2009 at 03:19:43AM +0200, Andi Kleen wrote:
> > > > There is no real rush AFAIKS to fix this one single pagecache site
> > > > while we have problems with slab allocators and all other unaudited
> > > > places that nonatomically modify page flags with an elevated
> > > 
> > > hwpoison ignores slab pages.
> > 
> > "ignores" them *after* it has already written to page flags?
> > By that time it's too late.
> 
> Yes, currently the page lock comes first. The only exception 
> is for page count == 0 pages. I suppose we could move the slab check
> up, but then it only helps when slab is set.

Yes, so it misses other potential non-atomic page flags manipulations.

 
> So if you make slab use refcount == 0 pages that would help.

Yes it would help here and also help with the pagecache part too,
and most other cases I suspect. I have some patches to do this at
home so I'll post them when I get back.

 
> > Well it's fundamentally badly buggy, rare or not. We could avoid
> 
> Let's put it like this -- any access to the poisoned cache lines
> in that page will trigger a panic anyways.

Well yes, although maybe people who care about this feature will
care more about having a reliable panic than introducing a
random data corruption. I guess the chance of an ecc failure
combined with a chance the race window hits could be some orders
of magnitude less likely than other sources of bugs ;) but still
I don't like using that argument to allow known bugs -- it leads
to interesting things if we take it to a conclusion.

--
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/