Re: [PATCH] x86: constrain sub-page access length in mmio_ro_emulated_write()

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

On Wed, Apr 23, 2025 at 02:58:28PM +0200, Jan Beulich wrote:
> On 23.04.2025 11:07, Roger Pau Monné wrote:
> > On Wed, Apr 23, 2025 at 10:43:56AM +0200, Jan Beulich wrote:
> >> Without doing so we could trigger the ASSERT_UNREACHABLE() in
> >> subpage_mmio_write_emulate(). A comment there actually says this
> >> validation would already have been done ...
> >>
> >> Fixes: 8847d6e23f97 ("x86/mm: add API for marking only part of a MMIO page 
> >> read only")
> >> Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>
> >> ---
> >> Alternatively we could drop comment and assertion from
> >> subpage_mmio_write_emulate().
> > 
> > I think I prefer this as it fits better with my patch to unify the
> > open-coded MMIO accessors, which does have an ASSERT_UNREACHABLE() for
> > unhandled sizes.  The return there is anyway too late IMO, as we have
> > possibly already mapped the page when there's no need for it.
> 
> FTAOD with "this" you mean the patch as is, not the alternative?

Yes, sorry, "this" => "this patch".

> >> --- a/xen/arch/x86/mm.c
> >> +++ b/xen/arch/x86/mm.c
> >> @@ -5195,8 +5195,9 @@ int cf_check mmio_ro_emulated_write(
> >>          return X86EMUL_UNHANDLEABLE;
> >>      }
> >>  
> >> -    subpage_mmio_write_emulate(mmio_ro_ctxt->mfn, PAGE_OFFSET(offset),
> >> -                               p_data, bytes);
> >> +    if ( bytes <= 8 )
> >> +        subpage_mmio_write_emulate(mmio_ro_ctxt->mfn, PAGE_OFFSET(offset),
> >> +                                   p_data, bytes);
> > 
> > Should we print a debug message here saying the write is possibly
> > unhandled due to the access size if subpage r/o is enabled?
> > 
> > You might want to re-use the subpage_ro_active() I introduce to limit
> > the printing of the message.
> 
> That would be too broad for my taste. I've used subpage_mmio_find_page()

Hm, yes, that's likely more expensive, but certainly more accurate.
Given the context here the extra logic doesn't matter much.

Thanks, Roger.