Re: [PATCH] x86/mm: Do not allow non-MAP_FIXED mapping across DEFAULT_MAP_WINDOW border

[Kirill A. Shutemov]

On Mon, Nov 13, 2017 at 08:14:54PM +0100, Thomas Gleixner wrote:
> On Mon, 13 Nov 2017, Thomas Gleixner wrote:
> > On Mon, 13 Nov 2017, Kirill A. Shutemov wrote:
> > 
> > > On Mon, Nov 13, 2017 at 04:43:26PM +0100, Thomas Gleixner wrote:
> > > > On Tue, 7 Nov 2017, Kirill A. Shutemov wrote:
> > > > 
> > > > > In case of 5-level paging, we don't put any mapping above 47-bit, unless
> > > > > userspace explicitly asked for it.
> > > > > 
> > > > > Userspace can ask for allocation from full address space by specifying
> > > > > hint address above 47-bit.
> > > > > 
> > > > > Nicholas noticed that current implementation violates this interface:
> > > > > we can get vma partly in high addresses if we ask for a mapping at very
> > > > > end of 47-bit address space.
> > > > > 
> > > > > Let's make sure that, when consider hint address for non-MAP_FIXED
> > > > > mapping, start and end of resulting vma are on the same side of 47-bit
> > > > > border.
> > > > 
> > > > What happens for mappings with MAP_FIXED which cross the border?
> > > 
> > > It will succeed with 5-level paging.
> > 
> > And why is this allowed?
> > 
> > > It should be safe as with 4-level paging such request would fail and it's
> > > reasonable to expect that userspace is not relying on the failure to
> > > function properly.
> > 
> > Huch?
> > 
> > The first rule when looking at user space is that is broken or
> > hostile. Reasonable and user space are mutually exclusive.
> 
> Aside of that in case of get_unmapped_area:
> 
> If va_unmapped_area() fails, then the address and the len which caused the
> overlap check to trigger are handed in to arch_get_unmapped_area(), which
> again can create an invalid mapping if I'm not missing something.
> 
> If mappings which overlap the boundary are invalid then we have to make
> sure at all ends that they wont happen.

They are not invalid.

The patch tries to address following theoretical issue:

We have an application that tries, for some reason, to allocate memory
with mmap(addr), without MAP_FIXED, where addr is near the borderline of
47-bit address space and addr+len is above the border.

On 4-level paging machine this request would succeed, but the address will
always be within 47-bit VA -- cannot allocate by hint address, ignore it.

If the application cannot handle high address this might be an issue on
5-level paging machine as such call would succeed *and* allocate memory by
the specified hint address. In this case part of the mapping would be
above the border line and may lead to misbehaviour.

I hope this makes any sense :)

-- 
 Kirill A. Shutemov