From: Tycho Andersen <tycho@docker.com>
To: Dave Hansen <dave.hansen@intel.com>
Cc: linux-kernel@vger.kernel.org, linux-mm@kvack.org,
kernel-hardening@lists.openwall.com,
Marco Benatto <marco.antonio.780@gmail.com>,
Juerg Haefliger <juerg.haefliger@canonical.com>,
x86@kernel.org
Subject: Re: [PATCH v6 03/11] mm, x86: Add support for eXclusive Page Frame Ownership (XPFO)
Date: Wed, 20 Sep 2017 18:09:01 -0600 [thread overview]
Message-ID: <20170921000901.v7zo4g5edhqqfabm@docker> (raw)
In-Reply-To: <97475308-1f3d-ea91-5647-39231f3b40e5@intel.com>
On Wed, Sep 20, 2017 at 04:21:15PM -0700, Dave Hansen wrote:
> On 09/20/2017 03:34 PM, Tycho Andersen wrote:
> >> I really have to wonder whether there are better ret2dir defenses than
> >> this. The allocator just seems like the *wrong* place to be doing this
> >> because it's such a hot path.
> >
> > This might be crazy, but what if we defer flushing of the kernel
> > ranges until just before we return to userspace? We'd still manipulate
> > the prot/xpfo bits for the pages, but then just keep a list of which
> > ranges need to be flushed, and do the right thing before we return.
> > This leaves a little window between the actual allocation and the
> > flush, but userspace would need another thread in its threadgroup to
> > predict the next allocation, write the bad stuff there, and do the
> > exploit all in that window.
>
> I think the common case is still that you enter the kernel, allocate a
> single page (or very few) and then exit. So, you don't really reduce
> the total number of flushes.
>
> Just think of this in terms of IPIs to do the remote TLB flushes. A CPU
> can do roughly 1 million page faults and allocations a second. Say you
> have a 2-socket x 28-core x 2 hyperthead system = 112 CPU threads.
> That's 111M IPI interrupts/second, just for the TLB flushes, *ON* *EACH*
> *CPU*.
Since we only need to flush when something switches from a userspace
to a kernel page or back, hopefully it's not this bad, but point
taken.
> I think the only thing that will really help here is if you batch the
> allocations. For instance, you could make sure that the per-cpu-pageset
> lists always contain either all kernel or all user data. Then remap the
> entire list at once and do a single flush after the entire list is consumed.
Just so I understand, the idea would be that we only flush when the
type of allocation alternates, so:
kmalloc(..., GFP_KERNEL);
kmalloc(..., GFP_KERNEL);
/* remap+flush here */
kmalloc(..., GFP_HIGHUSER);
/* remap+flush here */
kmalloc(..., GFP_KERNEL);
?
Tycho
next prev parent reply other threads:[~2017-09-21 0:09 UTC|newest]
Thread overview: 76+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-09-07 17:35 [PATCH v6 00/11] Add support for eXclusive Page Frame Ownership Tycho Andersen
2017-09-07 17:35 ` [PATCH v6 01/11] mm: add MAP_HUGETLB support to vm_mmap Tycho Andersen
2017-09-08 7:42 ` Christoph Hellwig
2017-09-07 17:36 ` [PATCH v6 02/11] x86: always set IF before oopsing from page fault Tycho Andersen
2017-09-07 17:36 ` [PATCH v6 03/11] mm, x86: Add support for eXclusive Page Frame Ownership (XPFO) Tycho Andersen
2017-09-07 18:33 ` Ralph Campbell
2017-09-07 18:50 ` Tycho Andersen
2017-09-08 7:51 ` Christoph Hellwig
2017-09-08 14:58 ` Tycho Andersen
2017-09-09 15:35 ` Laura Abbott
2017-09-11 15:03 ` Tycho Andersen
2017-09-11 7:24 ` Yisheng Xie
2017-09-11 14:50 ` Tycho Andersen
2017-09-11 16:03 ` Juerg Haefliger
2017-09-11 16:59 ` Tycho Andersen
2017-09-12 8:05 ` Yisheng Xie
2017-09-12 14:36 ` Tycho Andersen
2017-09-12 18:13 ` Tycho Andersen
2017-09-14 6:15 ` Yisheng Xie
2017-09-20 23:46 ` Dave Hansen
2017-09-21 0:02 ` Tycho Andersen
2017-09-21 0:04 ` Dave Hansen
2017-09-11 18:32 ` Tycho Andersen
2017-09-11 21:54 ` Marco Benatto
2017-09-20 15:48 ` Dave Hansen
2017-09-20 22:34 ` Tycho Andersen
2017-09-20 23:21 ` Dave Hansen
2017-09-21 0:09 ` Tycho Andersen [this message]
2017-09-21 0:27 ` Dave Hansen
2017-09-21 1:37 ` Tycho Andersen
2017-11-10 1:09 ` Tycho Andersen
2017-11-13 22:20 ` Dave Hansen
2017-11-13 22:46 ` Dave Hansen
2017-11-15 0:33 ` [kernel-hardening] " Tycho Andersen
2017-11-15 0:37 ` Dave Hansen
2017-11-15 0:42 ` Tycho Andersen
2017-11-15 3:44 ` Matthew Wilcox
2017-11-15 7:00 ` Dave Hansen
2017-11-15 14:58 ` Matthew Wilcox
2017-11-15 16:20 ` [kernel-hardening] " Tycho Andersen
2017-11-15 21:34 ` Matthew Wilcox
2017-09-21 0:03 ` Dave Hansen
2017-09-21 0:28 ` Dave Hansen
2017-09-21 1:04 ` Tycho Andersen
2017-09-07 17:36 ` [PATCH v6 04/11] swiotlb: Map the buffer if it was unmapped by XPFO Tycho Andersen
2017-09-07 18:10 ` Christoph Hellwig
2017-09-07 18:44 ` Tycho Andersen
2017-09-08 7:13 ` Christoph Hellwig
2017-09-07 17:36 ` [PATCH v6 05/11] arm64/mm: Add support for XPFO Tycho Andersen
2017-09-08 7:53 ` Christoph Hellwig
2017-09-08 17:24 ` Tycho Andersen
2017-09-14 10:41 ` Julien Grall
2017-09-14 11:29 ` Juergen Gross
2017-09-14 18:22 ` [kernel-hardening] " Mark Rutland
2017-09-18 21:27 ` Tycho Andersen
2017-09-07 17:36 ` [PATCH v6 06/11] xpfo: add primitives for mapping underlying memory Tycho Andersen
2017-09-07 17:36 ` [PATCH v6 07/11] arm64/mm, xpfo: temporarily map dcache regions Tycho Andersen
2017-09-14 18:25 ` Mark Rutland
2017-09-18 21:29 ` Tycho Andersen
2017-09-07 17:36 ` [PATCH v6 08/11] arm64/mm: Add support for XPFO to swiotlb Tycho Andersen
2017-09-07 17:36 ` [PATCH v6 09/11] arm64/mm: disable section/contiguous mappings if XPFO is enabled Tycho Andersen
2017-09-09 15:38 ` Laura Abbott
2017-09-07 17:36 ` [PATCH v6 10/11] mm: add a user_virt_to_phys symbol Tycho Andersen
2017-09-08 7:55 ` Christoph Hellwig
2017-09-08 15:44 ` Kees Cook
2017-09-11 7:36 ` Christoph Hellwig
2017-09-14 18:34 ` [kernel-hardening] " Mark Rutland
2017-09-18 20:56 ` Tycho Andersen
2017-09-07 17:36 ` [PATCH v6 11/11] lkdtm: Add test for XPFO Tycho Andersen
2017-09-07 19:08 ` Kees Cook
2017-09-10 0:57 ` kbuild test robot
2017-09-11 10:34 ` [PATCH v6 00/11] Add support for eXclusive Page Frame Ownership Yisheng Xie
2017-09-11 15:02 ` Tycho Andersen
2017-09-12 7:07 ` Yisheng Xie
2017-09-12 7:40 ` Juerg Haefliger
2017-09-12 8:11 ` Yisheng Xie
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20170921000901.v7zo4g5edhqqfabm@docker \
--to=tycho@docker.com \
--cc=dave.hansen@intel.com \
--cc=juerg.haefliger@canonical.com \
--cc=kernel-hardening@lists.openwall.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=marco.antonio.780@gmail.com \
--cc=x86@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).