Re: [PATCH 7/7] VT-d: move {,un}map_vtd_domain_page()

["Roger Pau Monné" <roger.pau@citrix.com>]

On Mon, Feb 12, 2024 at 10:46:55AM +0100, Jan Beulich wrote:
> On 09.02.2024 09:39, Roger Pau Monné wrote:
> > On Mon, Feb 05, 2024 at 02:57:30PM +0100, Jan Beulich wrote:
> >> ..., thus allowing them to become static. There's nothing x86-specific
> >> about these functions anyway.
> >>
> >> Since only the "iommu_inclusive_mapping" parameter declaration would be
> >> left in the file, move that as well. There's nothing VT-d specific about
> >> it (anymore?): "dom0-iommu=map-inclusive" is similarly generic, and
> >> documentation also doesn't say anything.
> > 
> > Hm, I guess documentation should at least say that
> > iommu_inclusive_mapping is x86 specific, because it's not parsed on
> > Arm and hence might give the wrong impression that it's actually
> > acknowledged there.
> 
> In v2 I'm adding "(x86)" there.

Maybe part of this should be done as a separate patch so it can be
backported to stable branches?

> >> Signed-off-by: Jan Beulich <jbeulich@suse.com>
> > 
> > Acked-by: Roger Pau Monné <roger.pau@citrix.com>
> 
> Thanks.
> 
> > Albeit I think it would be better to put the parsing in generic
> > iommu.c, so that the option gets parsed on Arm and
> > arch_iommu_hwdom_init() can print a warning message about it not
> > supported on Arm.
> 
> Hmm, I would have considered doing things the other way around - make
> that part of parsing in parse_dom0_iommu_param() x86-only. I would
> feel odd to introduce an option to Arm, just to be able to report
> that it's unsupported. The more when generic option parsing code will
> already log unrecognized options (sadly such log messages aren't seen
> in the serial log, for being issued too early). But let's ask Arm
> folks what they'd prefer, by adding all of them to To:.

FWIW, that (moving part of the parsing of parse_dom0_iommu_param() to
x86-only code) would also be fine.  As long as both
dom0-iommu=map-inclusive and iommu_inclusive_mapping are handled
equally and the documentation is updated to reflect that.

Thanks, Roger.