linux-kernel - Re: [PATCH] kvm/x86/mmu: use the correct inherited permissions to get shadow 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:   Tue, 1 Dec 2020 09:31:08 +0800
From:   Lai Jiangshan <jiangshanlai@...il.com>
To:     Sean Christopherson <seanjc@...gle.com>
Cc:     Paolo Bonzini <pbonzini@...hat.com>,
        LKML <linux-kernel@...r.kernel.org>, kvm@...r.kernel.org,
        Lai Jiangshan <laijs@...ux.alibaba.com>,
        Jonathan Corbet <corbet@....net>,
        Sean Christopherson <sean.j.christopherson@...el.com>,
        Vitaly Kuznetsov <vkuznets@...hat.com>,
        Wanpeng Li <wanpengli@...cent.com>,
        Jim Mattson <jmattson@...gle.com>,
        Joerg Roedel <joro@...tes.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
        X86 ML <x86@...nel.org>, "H. Peter Anvin" <hpa@...or.com>,
        Avi Kivity <avi@...ranet.com>, linux-doc@...r.kernel.org
Subject: Re: [PATCH] kvm/x86/mmu: use the correct inherited permissions to get
 shadow page

On Tue, Dec 1, 2020 at 1:41 AM Sean Christopherson <seanjc@...gle.com> wrote:

>
> Hmm, yes, KVM would incorrectly handle this scenario.  But, the proposed patch
> would not address the issue as KVM always maps non-leaf shadow pages with full
> access permissions.
>

Is it possible to exactly copy the access permissions from the guest
for non-leaf
shadow pages? Any protection from hypervisor (such as dirty track,
rmap_write_protect)
can only play on the leaf shadow ptes.

> Can we have a testcase in kvm-unit-tests?  It's okay of course if it
> only fails with ept=0.

Yes, it may have a flaw with ept=0. I don't get what "It's okay of course"
means. Is it related to kvm-unit-tests? Or no cloud provider uses
ept=0?