Re: [PATCH v5 01/11] xen/arm: xc_domain_ioport_permission(..) not supported on ARM.

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

On Tue, Oct 12, 2021 at 04:15:20PM +0000, Bertrand Marquis wrote:
> Hi Ian,
> 
> > On 12 Oct 2021, at 15:53, Ian Jackson <iwj@xxxxxxxxxxxxxx> wrote:
> > 
> > Bertrand Marquis writes ("Re: [PATCH v5 01/11] xen/arm: 
> > xc_domain_ioport_permission(..) not supported on ARM."):
> >> So there is no comment remaining on this patch ?
> > 
> > I have been following this thread.  I think I have a notion of why
> > this is needed but I'm not 100% clear on it.  The commit message says
> > this:
> > 
> >> ARM architecture does not implement I/O ports. Ignore this call on ARM
> >> to avoid the overhead of making a hypercall just for Xen to return
> >> -ENOSYS.
> > 
> > which implies it's a performance improvement.  But the change also
> > suppresses an error return, so this commit message is false.  I think
> > that the thread has concluded something different, but it should be
> > explained in the commit message.  The purpose of a commit message is
> > precisely to capture the kind of considerations and discussion that
> > occurred in this thread.
> 
> I can add something in the commit message about the fact that we improve
> performance and prevent to do a call that is and will not be supported in Xen.

IMO it would be good to modify the commit message so it covers the
fact that the emulated host bridge on Arm does not advertise IO port
support, so the guest is capable of realizing IO BARs are not
supported.

Otherwise it seems like the toolstack is ignoring a failure which
could cause a device to malfunction when passed though (which is still
the case, but the guest will be able to notice).

Thanks, Roger.