Re: [PATCH] x86/hvm: don't treat MMIO pages as special ones regarding cache attributes

[Roger Pau Monné <roger.pau@xxxxxxxxxx>]

On Thu, Sep 10, 2020 at 01:28:44PM +0200, Jan Beulich wrote:
> On 10.09.2020 13:05, Roger Pau Monné wrote:
> > It's still not clear to me what option would be better: modify
> > share_xen_page_with_guest to not mark pages as Xen heap, or implement
> > something different to assign MMIO pages to dom_io without setting
> > the Xen heap flag.
> 
> static void __init share_io_page(struct page_info *page)
> {
>     set_gpfn_from_mfn(mfn_x(page_to_mfn(page)), INVALID_M2P_ENTRY);
> 
>     /* The incremented type count pins as writable. */
>     page->u.inuse.type_info = PGT_writable_page | PGT_validated | 1;
> 
>     page_set_owner(page, dom_io);
> 
>     page->count_info |= PGC_allocated | 1;
> }
> 
> is of course much shorter than share_xen_page_with_guest(), but
> I'm nevertheless uncertain whether simply making conditional
> the setting of PGC_xen_heap there isn't the easier route. Of
> course, not pointlessly acquiring and releasing a lock has its
> own appeal.

I've went over the existing is_special_page users and I think it's
fine for MMIO regions to not be marked as special pages.

Will send a new patch.

Roger.