Re: [PATCH 31/32] x86, pkeys: execute-only support

[Kees Cook]

On Mon, Dec 14, 2015 at 3:39 PM, Andy Lutomirski <luto@xxxxxxxxxxxxxx> wrote:
> On Mon, Dec 14, 2015 at 3:37 PM, Dave Hansen <dave@xxxxxxxx> wrote:
>> On 12/14/2015 12:05 PM, Kees Cook wrote:
>>> On Mon, Dec 14, 2015 at 11:06 AM, Dave Hansen <dave@xxxxxxxx> wrote:
>>>> > From: Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx>
>>>> > Protection keys provide new page-based protection in hardware.
>>>> > But, they have an interesting attribute: they only affect data
>>>> > accesses and never affect instruction fetches.  That means that
>>>> > if we set up some memory which is set as "access-disabled" via
>>>> > protection keys, we can still execute from it.
>> ...
>>>> > I haven't found any userspace that does this today.
>>> To realistically take advantage of this, it sounds like the linker
>>> would need to know to keep bss and data page-aligned away from text,
>>> and then set text to PROT_EXEC only?
>>>
>>> Do you have any example linker scripts for this?
>>
>> Nope.  My linker-fu is weak.
>>
>> Can we even depend on the linker by itself?  Even if the sections were
>> marked --x, we can't actually use them with those permissions unless we
>> have protection keys.
>>
>> Do we need some special tag on the section to tell the linker to map it
>> as --x under some conditions and r-x for others?
>>
>
> Why?  Wouldn't --x just end up acting like r-x if PKRU is absent?

Good point!

What was the mixed section problem that came up before? I realize I
said "bss/data" before, but that's not right: those are already
page-aligned since they're writable, and rodata would be too, only
non-executable. What was the case of memory that needed to be
readable? ENEEDCOFFEE.

-Kees

-- 
Kees Cook
Chrome OS & Brillo Security
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/