Re: [PATCH V2 09/23] xen/dm: Make x86's DM feature common

[Oleksandr <olekstysh@xxxxxxxxx>]

On 12.11.20 13:32, Jan Beulich wrote:

Hi Jan

> On 15.10.2020 18:44, Oleksandr Tyshchenko wrote:
> > From: Julien Grall <julien.grall@xxxxxxx>
> >
> > As a lot of x86 code can be re-used on Arm later on, this patch
> > splits devicemodel support into common and arch specific parts.
> >
> > The common DM feature is supposed to be built with IOREQ_SERVER
> > option enabled (as well as the IOREQ feature), which is selected
> > for x86's config HVM for now.
> Did you consider doing it the other way around? It would seem
> more natural to have the top level dm-op handling arch-specific
> and call into e.g. ioreq_server_dm_op() for otherwise unhandled
> ops, just like e.g. do_domctl() calls into iommu_do_domctl()
> (indirectly via arch_do_domctl()). I ask because in the long
> run I expect the ioreq server sub-ops to only be a small part
> of the overall set of dm-ops; already now it's 7 out of 18 if
> I got the counting right.
>
> This would then also leave compat_dm_op() in x86 code.
>
> But yes, there are also downsides with this alternative.

No, I didn't consider. I separated the proposed DM changes from Julien's 
patch without modifying the logic.
My changes on top (except rebasing of course) are updating XSM code and 
introducing a xen/dm.h for definitions.
--
Regards,

Oleksandr Tyshchenko