linux-kernel - Re: Xen MMU's requirement to pin pages RO and initial_memory

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: <alpine.DEB.2.00.1105131650550.8972@kaball-desktop>
Date:	Mon, 16 May 2011 11:23:14 +0100
From:	Stefano Stabellini <stefano.stabellini@...citrix.com>
To:	Konrad Rzeszutek Wilk <konrad.wilk@...cle.com>
CC:	"yinghai@...nel.org" <yinghai@...nel.org>,
	"jeremy@...p.org" <jeremy@...p.org>,
	Stefano Stabellini <Stefano.Stabellini@...citrix.com>,
	"hpa@...or.com" <hpa@...or.com>,
	"hpa@...ux.intel.com" <hpa@...ux.intel.com>,
	Ian Campbell <Ian.Campbell@...citrix.com>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"xen-devel@...ts.xensource.com" <xen-devel@...ts.xensource.com>
Subject: Re: Xen MMU's requirement to pin pages RO and
 initial_memory_mapping.

On Fri, 13 May 2011, Konrad Rzeszutek Wilk wrote:
> aka, remove the hack added by git commit 609cfda586c7fe3e5d1a02c51edb587506294167
> (Merge branch 'stable/bug-fixes-for-rc5' of git://git.kernel.org/../xen)
> 
> One idea that is on the table was proposed by Yinghai:
> "Xen should set RAM for page-table to RO after init_memory mapping."
> 
> In other words, don't do the magic 'mask_rw_pte' when set_pte is called.
> (and don't do the calls to make_lowmem_page_readonly when allocating
> the PTE table, nor PMD, nor PUD) - I think?
> 
> But instead do:
>  a). when we load the cr3? We could go through the whole pagetable and
>      set the RO as we need?
>  b). when we are finished with the creation of a page table? So similary 
>      to the point above - don't set the RO on the pages until we have completed
>      the full creation of the page.
>  c). when post_allocator_start is called?

I think c) is too late because there might be allocations or at least
memblock allocations after init_memory_mapping and before
post_allocator_start.


>  d). other ideas?
> 
> But I vaguelly recall that we are using the page table as we are adding in the
> entries. And that we are pinning them as well. Perhaps the trigger to scan the
> pagetable and set them to RO should be done .. at what point? When the PUD/PMD
> allocations are done? And when PUD/PMD are set?
> 

Setting the pagetable pages RO after init_memory_mapping is difficult
because pagetable pages have to be set RO before becoming pagetable
pages.
They become pagetable pages when:

- they are explicitly pinned by pin_pagetable_pfn

- they are hooked into the current pagetable

Like you wrote, considering that the x86_64 version of
kernel_physical_mapping_init hooks the pagetable pages into the
currently used pagetable, it wouldn't be possible to mark the pagetable
pages RO after init_memory_mapping.
--
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/