linux-kernel - Re: [RFC PATCH 1/2] x86: Don't let pgprot_modify() change the page encryption bit

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, 5 Sep 2019 18:29:25 +0200
From:   Thomas Hellström (VMware) 
        <thomas_os@...pmail.org>
To:     Dave Hansen <dave.hansen@...el.com>, linux-kernel@...r.kernel.org,
        x86@...nel.org, pv-drivers@...are.com
Cc:     Thomas Hellstrom <thellstrom@...are.com>,
        Dave Hansen <dave.hansen@...ux.intel.com>,
        Andy Lutomirski <luto@...nel.org>,
        Peter Zijlstra <peterz@...radead.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
        "H. Peter Anvin" <hpa@...or.com>,
        Christoph Hellwig <hch@...radead.org>,
        Christian König <christian.koenig@....com>,
        Marek Szyprowski <m.szyprowski@...sung.com>,
        Tom Lendacky <thomas.lendacky@....com>,
        "Shutemov, Kirill" <kirill.shutemov@...el.com>
Subject: Re: [RFC PATCH 1/2] x86: Don't let pgprot_modify() change the page
 encryption bit

On 9/5/19 5:59 PM, Dave Hansen wrote:
> On 9/5/19 8:21 AM, Thomas Hellström (VMware) wrote:
>>>>    #define pgprot_modify pgprot_modify
>>>>    static inline pgprot_t pgprot_modify(pgprot_t oldprot, pgprot_t
>>>> newprot)
>>>>    {
>>>> -    pgprotval_t preservebits = pgprot_val(oldprot) & _PAGE_CHG_MASK;
>>>> -    pgprotval_t addbits = pgprot_val(newprot);
>>>> +    pgprotval_t preservebits = pgprot_val(oldprot) &
>>>> +        (_PAGE_CHG_MASK | sme_me_mask);
>>>> +    pgprotval_t addbits = pgprot_val(newprot) & ~sme_me_mask;
>>>>        return __pgprot(preservebits | addbits);
>>>>    }
>>> _PAGE_CHG_MASK is claiming similar functionality about preserving bits
>>> when changing PTEs:
> ...
>>>> #define _PAGE_CHG_MASK  (PTE_PFN_MASK | _PAGE_PCD | _PAGE_PWT
>>>> |         \
>>>>                            _PAGE_SPECIAL | _PAGE_ACCESSED |
>>>> _PAGE_DIRTY | \
>>>>                            _PAGE_SOFT_DIRTY | _PAGE_DEVMAP)
>>> This makes me wonder if we should be including sme_me_mask in
>>> _PAGE_CHG_MASK (logically).
>> I was thinking the same. But what confuses me is that addbits isn't
>> masked with ~_PAGE_CHG_MASK, which is needed for sme_me_mask, since the
>> problem otherwise is typically that the encryption bit is incorrectly
>> set in addbits. I wonder whether it's an optimization or intentional.
> I think there's a built-in assumption that 'newprot' won't have any of
> the _PAGE_CHG_MASK bits set.  That makes sense because there are no
> protection bits in the mask.  But, the code certainly doesn't enforce that.
>
> Are you seeing 'sme_me_mask' bits set in 'newprot'?

Yes. AFAIK it's only one bit, and typically always set.

/Thomas