Re: [PATCH v34 10/24] mm: Add vm_ops->mprotect()

[Sean Christopherson <sean.j.christopherson@intel.com>]

On Tue, Jul 07, 2020 at 04:14:24AM +0100, Matthew Wilcox wrote:
> On Tue, Jul 07, 2020 at 06:01:50AM +0300, Jarkko Sakkinen wrote:
> > +++ b/mm/mprotect.c
> > @@ -603,13 +603,20 @@ static int do_mprotect_pkey(unsigned long start, size_t len,
> >  			goto out;
> >  		}
> >  
> > +		tmp = vma->vm_end;
> > +		if (tmp > end)
> > +			tmp = end;
> > +
> >  		error = security_file_mprotect(vma, reqprot, prot);
> >  		if (error)
> >  			goto out;
> >  
> > -		tmp = vma->vm_end;
> > -		if (tmp > end)
> > -			tmp = end;
> 
> You don't need to move this any more, right?

Ya, I was typing up a longer version, you beat me to the punch...

The calculation of 'tmp' doesn't need to be moved.  The previous incarnation
only moved it so that that 'tmp would be available for .may_mprotect().

The only reason  .may_mprotect() was placed before security_file_mprotect()
was to avoid triggering LSM errors that would have been blocked by
.may_mprotect(), but that justification is no longer relevant as the
proposed SGX LSM hooks obviously didn't worm their way into this series.

> > +		if (vma->vm_ops && vma->vm_ops->mprotect) {
> > +			error = vma->vm_ops->mprotect(vma, nstart, tmp, prot);
> > +			if (error)
> > +				goto out;
> > +		}

Based on "... and then the vma owner can do whatever it needs to before
calling mprotect_fixup(), which is already not static", my interpretation
is that Matthew's intent was to do:

		if (vma->vm_ops && vma->vm_ops->mprotect)
			error =  = vma->vm_ops->mprotect(vma, nstart, tmp, prot);
		else
			error = mprotect_fixup(vma, &prev, nstart, tmp, newflags);
		if (error)
			goto out;

i.e. make .mprotect() a full replacement as opposed to a prereq hook.
		
> > +
> >  		error = mprotect_fixup(vma, &prev, nstart, tmp, newflags);
> >  		if (error)
> >  			goto out;
> > -- 
> > 2.25.1
> >