Re: [Xen-devel] [PATCH v2 15/27] ARM: vITS: introduce translation table walks

[Julien Grall <julien.grall@xxxxxxx>]

Hi Andre,

On 03/04/17 19:25, Andre Przywara wrote:
> On 24/03/17 13:00, Julien Grall wrote:
> > On 03/16/2017 11:20 AM, Andre Przywara wrote:
> > > + */
> > > +bool read_itte(struct virt_its *its, uint32_t devid, uint32_t evid,
> > > +               struct vcpu **vcpu, uint32_t *vlpi)
> > > +{
> > > +    struct vits_itte *itte;
> > > +    int collid;
> > > +    uint32_t _vlpi;
> > > +    struct vcpu *_vcpu;
> > > +
> > > +    spin_lock(&its->its_lock);
> > Do we expect multiple vCPU calling read_itte at the same time? This
> > needs to be clarify in the comments because this current function is not
> > scalable.
> We need this lock here because this protects our data structures.
> A guest locks accesses to the ITS command queue anyway (otherwrite
> multiple VCPUs would race for CWRITER and CREADR), so I don't see a real
> problem. And this lock is pre ITS, not per guest.
Fair point, please ignore my request then.

Cheers,

--
Julien Grall

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel