Re: [PATCH] arm64/xen: fix xen-swiotlb cache flushing

[Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>]

[full quote deleted, please take a little more care when quoting]

On Fri, Jan 18, 2019 at 04:44:23PM -0800, Stefano Stabellini wrote:
> >  #ifdef CONFIG_XEN
> > -	if (xen_initial_domain()) {
> > -		dev->archdata.dev_dma_ops = dev->dma_ops;
> > +	if (xen_initial_domain())
> >  		dev->dma_ops = xen_dma_ops;
> > -	}
> >  #endif
> >  }
> 
> This is an optional suggestion, but it would be nice to add a check on
> dev->dma_ops being unset here, something like:
> 
>   #ifdef CONFIG_XEN
>   if (xen_initial_domain()) {
>       if (dev->dma_ops != NULL)
>           warning/error
>       dev->dma_ops = xen_dma_ops;
>   }
> 
> Does it make sense?

Well, no such check existed before, so this probably should be a
separate patch if we care enough.  I have a series for 5.1 pending
that moves the IOMMU handling to the comment code which will make
the ops assginment a lot cleaner, and I guess I could fold such
a check in that.  Doing it now will just create churn as it would
have to get reworked anyway

> Reviewed-by: Stefano Stabellini <sstabellini-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>

Where should we pick this up?  I could pick it up through the dma-mapping
tree given that is where the problem is introduced, but the Xen or arm64
trees would also fit.