linux-kernel - Re: [PATCH 01/11] x86/mm: factor out pageattr _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]

Message-ID: <36637d85-d716-a2d4-189c-10ed209f4827@linux.intel.com>
Date:   Mon, 2 Apr 2018 13:23:04 -0700
From:   Dave Hansen <dave.hansen@...ux.intel.com>
To:     Linus Torvalds <torvalds@...ux-foundation.org>
Cc:     Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        linux-mm <linux-mm@...ck.org>,
        Andrea Arcangeli <aarcange@...hat.com>,
        Andrew Lutomirski <luto@...nel.org>,
        Kees Cook <keescook@...gle.com>,
        Hugh Dickins <hughd@...gle.com>,
        Jürgen Groß <jgross@...e.com>,
        the arch/x86 maintainers <x86@...nel.org>, namit@...are.com
Subject: Re: [PATCH 01/11] x86/mm: factor out pageattr _PAGE_GLOBAL setting

On 04/02/2018 10:52 AM, Linus Torvalds wrote:
> On Mon, Apr 2, 2018 at 10:27 AM, Dave Hansen
> <dave.hansen@...ux.intel.com> wrote:
>>
>> Aside: _PAGE_GLOBAL is ignored when CR4.PGE=1, so why do we
>> even go to the trouble of filtering it anywhere?
> 
> I'm assuming this is a typo, and you mean "when CR4.PGE=0".

Yes, that is a typo.

> The question you raise may be valid, but within the particular context
> of *this* patch it is not.

I thought it was relevant because I was asking myself: Why is it OK for
the (old) code to be doing this:

> -	if (pgprot_val(req_prot) & _PAGE_PRESENT)
> -		pgprot_val(req_prot) |= _PAGE_PSE | _PAGE_GLOBAL;

When _PAGE_GLOBAL is not supported.  This "Aside" got moved a bit away
from the comment, but I actually mean to refer to the comment that talks
about canon_pgprot():

>> canon_pgprot() will clear it if unsupported,
>> but we *always* set it.

and its use of __supported_pte_mask.

I'll redo the changelog a bit and hopefully capture all this along with
correcting the typo.