[PATCH v5 24/38] mm/vma: enforce that mm-managed mappings may not set VMA_IO_BIT

[Lorenzo Stoakes (ARM)]

On Mon, Oct 12, 2026 at 08:39:35PM +0200, David Hildenbrand (Arm) wrote:
> On 10/12/26 20:35, Lorenzo Stoakes (ARM) wrote:
> > On Mon, Oct 12, 2026 at 08:25:54PM +0200, David Hildenbrand (Arm) wrote:
> >> On 10/9/26 20:04, Lorenzo Stoakes (ARM) wrote:
> >>> It makes no sense for a mapping whose contents the kernel does not own to
> >>> specify that the range is MMIO.
> >>>
> >>> Prior to this patch, all in-tree drivers which did so have been updated
> >>> such that they are marked as not being mm-managed. The check WARNs and
> >>> fails the mmap for any out-of-tree driver that still sets VMA_IO_BIT
> >>> without a kernel mapping.
> >>>
> >>> No functional change intended for in-tree code.
> >>>
> >>> Reviewed-by: Zi Yan <ziy at nvidia.com>
> >>> Signed-off-by: Lorenzo Stoakes (ARM) <ljs at kernel.org>
> >>> ---
> >>>  mm/vma.c | 6 ++++++
> >>>  1 file changed, 6 insertions(+)
> >>>
> >>> diff --git a/mm/vma.c b/mm/vma.c
> >>> index 4f09fb98a2fd..51bdbe08bb02 100644
> >>> --- a/mm/vma.c
> >>> +++ b/mm/vma.c
> >>> @@ -2796,6 +2796,12 @@ static int mmap_validate_vma_flags(const vma_flags_t *flags)
> >>>  		return -EINVAL;
> >>>  #endif
> >>>
> >>> +	if (!vma_flags_test_any(flags, VMA_PFNMAP_BIT, VMA_MIXEDMAP_BIT)) {
> >>> +		/* mm-managed mappings may not set VMA_IO_BIT. */
> >>> +		if (WARN_ON_ONCE(vma_flags_test(flags, VMA_IO_BIT)))
> >>> +			return -EINVAL;
> >>
> >> We do have a helper (vma_flags_is_mm_managed()) but we're not using it because
> >> of the VM_IO situation ...
> >
> > That gets removed.
> >
> >>
> >> That's confusing ...
> >>
> >> Could we do something like testing vma_flags_is_mm_managed(), but removing the
> >> VM_IO flag from the checked flags?
> >
> > I mean we do that in a subsequent commit, I think even the next one?
> >
> >
> >>
> >> Then we can add a comment about VM_IO in the future getting remove from
> >> vma_flags_is_mm_managed() ...
> >
> > Yeah, but to do that the code would now be something like:
> >
> > 	vma_flags_t masked = flags;
> >
> > 	/*
> > 	  * vma_flags_is_mm_managed() tests VMA_IO_BIT, which will be removed
> > 	 * in the next commit.
> > 	 */
> > 	vma_flags_clear(&flags, VMA_IO_BIT);
> > 	if (!vma_flags_is_mm_managed(masked)) {
> > 		...
> > 	}
> >
> > Is that really worth it?
> >
> > I don't think there's really too much confusion intra-patch about that is there?
>
> I would do it as above (- the comment, as that is the next patch indeed) and
> only remove the vma_flags_clear() in the next patch.
>
> But in any case, you should spell out that this is only temporary in the patch
> description :)

And then there's a 'masked' variable that is unused?

You need to have two vma_flags_t variables to be able both clear VMA_IO_BIT
and use vma_flags_is_mm_managed().

So the change would have to be

+ 	vma_flags_t masked = *flags;
+
+ 	vma_flags_clear(&flags, VMA_IO_BIT);
+ 	if (!vma_flags_is_mm_managed(&masked)) {
+ 		...
+ 	}

Then:

- 	vma_flags_t masked = *flags;
-
- 	vma_flags_clear(&flags, VMA_IO_BIT);
- 	if (!vma_flags_is_mm_managed(&masked)) {
+ 	if (!vma_flags_is_mm_managed(flags)) {
 		...
 	}

Which doesn't really seem worth it for somebody who is confused between two
patches mid-series.

But if you feel strongly I can make the change like that?

--
Cheers, Lorenzo