Re: [PATCH] vfio: Depend on MMU

[Cornelia Huck <cohuck@redhat.com>]

On Mon, 8 Mar 2021 10:59:13 -0700
Alex Williamson <alex.williamson@redhat.com> wrote:

> On Fri, 5 Mar 2021 09:46:49 +0100
> Cornelia Huck <cohuck@redhat.com> wrote:
> 
> > On Thu, 4 Mar 2021 21:30:03 -0400
> > Jason Gunthorpe <jgg@nvidia.com> wrote:
> >   
> > > VFIO_IOMMU_TYPE1 does not compile with !MMU:
> > > 
> > > ../drivers/vfio/vfio_iommu_type1.c: In function 'follow_fault_pfn':
> > > ../drivers/vfio/vfio_iommu_type1.c:536:22: error: implicit declaration of function 'pte_write'; did you mean 'vfs_write'? [-Werror=implicit-function-declaration]
> > > 
> > > So require it.
> > > 
> > > Suggested-by: Cornelia Huck <cohuck@redhat.com>
> > > Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
> > > ---
> > >  drivers/vfio/Kconfig | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > diff --git a/drivers/vfio/Kconfig b/drivers/vfio/Kconfig
> > > index 90c0525b1e0cf4..67d0bf4efa1606 100644
> > > --- a/drivers/vfio/Kconfig
> > > +++ b/drivers/vfio/Kconfig
> > > @@ -22,7 +22,7 @@ config VFIO_VIRQFD
> > >  menuconfig VFIO
> > >  	tristate "VFIO Non-Privileged userspace driver framework"
> > >  	select IOMMU_API
> > > -	select VFIO_IOMMU_TYPE1 if (X86 || S390 || ARM || ARM64)
> > > +	select VFIO_IOMMU_TYPE1 if MMU && (X86 || S390 || ARM || ARM64)
> > >  	help
> > >  	  VFIO provides a framework for secure userspace device drivers.
> > >  	  See Documentation/driver-api/vfio.rst for more details.    
> > 
> > Actually, I'm wondering how much sense vfio makes on !MMU at all? (And
> > maybe just merge this with your patch that switches IOMMU_API from a
> > depend to a select, because that is the change that makes the MMU
> > dependency required?)  
> 
> We do have the no-iommu code in vfio, potentially it's useful for !MMU,
> I guess.  It seems a little arbitrary to remove it without a known
> breakage at this point.  Thanks,
> 
> Alex

Well, in practice, I think we had an implicit dependency on MMU before
(everything selecting IOMMU_API depended on MMU.) If we think !MMU
would be useful for the no-iommu use case, we can certainly restrict
the dependency to VFIO_IOMMU_TYPE1.