linux-kernel - Re: [PATCH v4 6/7] mtrr, x86: Clean up mtrr_type

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:	Wed, 06 May 2015 17:42:10 -0600
From:	Toshi Kani <toshi.kani@...com>
To:	Borislav Petkov <bp@...en8.de>
Cc:	akpm@...ux-foundation.org, hpa@...or.com, tglx@...utronix.de,
	mingo@...hat.com, linux-mm@...ck.org, x86@...nel.org,
	linux-kernel@...r.kernel.org, dave.hansen@...el.com,
	Elliott@...com, pebolle@...cali.nl
Subject: Re: [PATCH v4 6/7] mtrr, x86: Clean up mtrr_type_lookup()

On Thu, 2015-05-07 at 00:49 +0200, Borislav Petkov wrote:
> On Wed, May 06, 2015 at 10:00:30AM -0600, Toshi Kani wrote:
> > Ingo asked me to describe this info here in his review...
> 
> Ok.
> 
> > mtrr_type_lookup_fixed() checks the above conditions at entry, and
> > returns immediately with TYPE_INVALID.  I think it is safer to have such
> > checks in mtrr_type_lookup_fixed() in case there will be multiple
> > callers.
> 
> This is not what I mean - I mean to call mtrr_type_lookup_fixed() based
> on @start and not unconditionally, like you do.
> 
> And there most likely won't be multiple callers because we're phasing
> out MTRR use.
> 
> And even if there are, they better look at how this function is being
> called before calling it. Which I seriously doubt - it is a static
> function which you *just* came up with.

Well, creating mtrr_type_lookup_fixed() is one of the comments I had in
the previous code review.  Anyway, let me make sure if I understand your
comment correctly.  Do the following changes look right to you?

1) Change the caller responsible for the condition checks.

        if ((start < 0x100000) &&
            (mtrr_state.have_fixed) &&
            (mtrr_state.enabled & MTRR_STATE_MTRR_FIXED_ENABLED))
                return mtrr_type_lookup_fixed(start, end);

2) Delete the checks with mtrr_state in mtrr_type_lookup_fixed() as they
are done by the caller.  Keep the check with '(start >= 0x100000)' to
assure that the code handles the range [0xC0000 - 0xFFFFF] correctly.

static u8 mtrr_type_lookup_fixed(u64 start, u64 end)
{
        int idx;

        if (start >= 0x100000)
                 return MTRR_TYPE_INVALID;
 
-       if (!(mtrr_state.have_fixed) ||
-           !(mtrr_state.enabled & MTRR_STATE_MTRR_FIXED_ENABLED))
-               return MTRR_TYPE_INVALID;


> > Right, and there is more.  As the original code had comment "Just return
> > the type as per start", which I noticed that I had accidentally removed,
> > the code only returns the type of the start address.  The fixed ranges
> > have multiple entries with different types.  Hence, a given range may
> > overlap with multiple fixed entries.  I will restore the comment in the
> > function header to clarify this limitation.
> 
> Ok, let's cleanup this function first and then consider fixing other
> possible bugs which haven't been fixed since forever. Again, we might
> not even need to address them because we won't be using MTRRs once we
> switch to PAT completely, which is what Luis is working on.

Right.

Thanks,
-Toshi

--
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/