linux-kernel - Re: [PATCH v2 0/5] Add support for EPT execute only for nested hypervisors

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, 14 Jul 2016 14:52:32 -0400
From:	Bandan Das <bsd@...hat.com>
To:	Paolo Bonzini <pbonzini@...hat.com>
Cc:	kvm@...r.kernel.org, guangrong.xiao@...ux.intel.com,
	kernellwp@...il.com, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2 0/5] Add support for EPT execute only for nested hypervisors

Paolo Bonzini <pbonzini@...hat.com> writes:

> On 14/07/2016 19:38, Bandan Das wrote:
>> Paolo Bonzini <pbonzini@...hat.com> writes:
>> 
>>> On 13/07/2016 17:47, Bandan Das wrote:
>>>>>> I wanted to keep it the former way because "PT_PRESENT_MASK is equal to VMX_EPT_READABLE_MASK"
>>>>>> is an assumption all throughout. I wanted to use this section to catch mismatches.
>>>>>
>>>>> I think there's no such assumption anymore, actually.  Can you double
>>>>> check?  If there are any, that's where the BUILD_BUG_ON should be.
>>>>
>>>> What I meant is how they are the same bit.  is_shadow_present_pte() is probably one
>>>> and another one is link_shadow_page() which already has a BUILD_BUG_ON().
>>>
>>> You're right about link_shadow_page()!  We probably should change the
>>> PT_PRESENT_MASK to shadow_present_mask there (and then readability in
>>> the EPT execonly case is still provided by shadow_user_mask).
>> 
>> Makes sense. Would you like a new version with that added or can that be a
>> separate patch ?
>
> I've already done it and pushed it to kvm/next. :)

Ah, thank you!

>>> For is_shadow_present_pte() you have removed it in patch 1 though.
>> 
>> Right. But the assumption is still that is_shadow_present_pte() works because
>> EPT_READABLE and PT_PRESENT are the same.
>
> is_shadow_present_pte() tests 0xFFFFFFFF, so it does not depend on bit 0
> alone, for neither EPT nor "normal" page tables.

Yeah... Let me rephrase, is_shadow_present_pte works because the assumption
is that both of the bits are in the first 32 bits :) You proved me wrong though,
this assumption does not mean a BUILD_BUG for the equal condition is required here.

Bandan

> Paolo