Re: Rethinking sigcontext's xfeatures slightly for PKRU's benefit?

[Andy Lutomirski <luto@amacapital.net>]

On Fri, Dec 18, 2015 at 3:08 PM, Linus Torvalds
<torvalds@linux-foundation.org> wrote:
> On Fri, Dec 18, 2015 at 2:28 PM, Andy Lutomirski <luto@amacapital.net> wrote:
>>
>> Apps that don't want to use the baseline_pkru mechanism could use
>> syscalls to claim ownership of protection keys but then manage them
>> purely with WRPKRU directly.  We could optionally disallow
>> mprotect_key on keys that weren't allocated in advance.
>>
>> Does that seem sane?
>
> So everything seems sane except for the need for that baseline_pkru.
>
> I'm not seeing why it couldn't just be a fixed value. Is there any
> real downside to it?

Yes, I think.  If I'm using protection keys to protect some critical
data structure (important stuff in shared memory, important memory
mapped files, pmem, etc), then I'll allocate a protection key and set
PKRU to deny writes.  The problem is that I really, really want writes
denied except when explicitly enabled in narrow regions of code that
use wrpkru to enable them, and I don't want an asynchronous signal
delivered in those narrow regions of code or newly cloned threads to
pick up the write-allow value.  So I want baseline_pkru to have the
deny writes entry.

I think I would do exactly this in my production code here if my
server supported it.  Some day...

Hrm.  We might also want an option to change pkru and/or baseline_pkru
in all threads in the current mm.  That's optional but it could be
handy.  Maybe it would be as simple as having the allocate-a-pkey call
have an option to set an initial baseline value and an option to
propagate that initial value to pre-existing threads.

--Andy