Re: [PATCH] mm: Add new vma flag VM_LOCAL_CPU

From: Boaz Harrosh
Date: Tue May 15 2018 - 13:38:05 EST

On 15/05/18 15:09, Peter Zijlstra wrote:
> On Tue, May 15, 2018 at 02:41:41PM +0300, Boaz Harrosh wrote:
>> On 15/05/18 14:11, Matthew Wilcox wrote:
>>> You're still thinking about this from the wrong perspective. If you
>>> were writing a program to attack this facility, how would you do it?
>>> It's not exactly hard to leak one pointer's worth of information.
>> That would be very hard. Because that program would:
>> - need to be root
>> - need to start and pretend it is zus Server with the all mount
>> thread thing, register new filesystem, grab some pmem devices.
>> - Mount the said filesystem on said pmem. Create core-pinned ZT threads
>> for all CPUs, start accepting IO.
>> - And only then it can start leaking the pointer and do bad things.
>> The bad things it can do to the application, not to the Kernel.
> No I think you can do bad things to the kernel at that point. Consider
> it populating the TLBs on the 'wrong' CPU by 'inadvertenly' touching
> 'random' memory.
> Then cause an invalidation and get the page re-used for kernel bits.
> Then access that page through the 'stale' TLB entry we still have on the
> 'wrong' CPU and corrupt kernel data.

Yes a BAD filesystem Server can do bad things I agree. But a filesystem can
do very bad things in any case. through the front door, No? and we trust
it with our data. So there is some trust we already put in a filesystem i think.

I will try to look at this deeper, see if I can actually enforce this policy.
Do you have any ideas? can I force page_faults on the other cores?

Thank you for looking