Re: [PATCH V1 03/16] xen/ioreq: Make x86's hvm_ioreq_needs_completion() common

[Oleksandr <olekstysh@xxxxxxxxx>]

On 14.09.20 17:59, Jan Beulich wrote:

Hi Jan

> On 10.09.2020 22:21, Oleksandr Tyshchenko wrote:
> > --- a/xen/include/xen/ioreq.h
> > +++ b/xen/include/xen/ioreq.h
> > @@ -35,6 +35,13 @@ static inline struct hvm_ioreq_server 
> > *get_ioreq_server(const struct domain *d,
> >       return GET_IOREQ_SERVER(d, id);
> >   }
> >   
> > +static inline bool hvm_ioreq_needs_completion(const ioreq_t *ioreq)
> > +{
> > +    return ioreq->state == STATE_IOREQ_READY &&
> > +           !ioreq->data_is_ptr &&
> > +           (ioreq->type != IOREQ_TYPE_PIO || ioreq->dir != IOREQ_WRITE);
> > +}
> While the PIO aspect has been discussed to some length, what about
> the data_is_ptr concept? I didn't think there were Arm insns fitting
> this? Instead I thought some other Arm-specific adjustments to the
> protocol might be needed. At which point the question of course would
> be in how far ioreq_t as a whole really fits Arm in its current shape.
I may mistake here but I don't think the "data_is_ptr" is supported.
It worth mentioning that on Arm, all the accesses to MMIO region will do 
a single memory access.
So we set "df" to 0 and "count" to 1. Other ioreq_t fields are in use.

--
Regards,

Oleksandr Tyshchenko