Re: [Xen-devel] [PATCH v4 7/8] dm_op: convert HVMOP_inject_trap and HVMOP_inject_msi

[Paul Durrant <Paul.Durrant@xxxxxxxxxx>]

> -----Original Message-----
> From: Andrew Cooper
> Sent: 20 January 2017 18:34
> To: Paul Durrant <Paul.Durrant@xxxxxxxxxx>; xen-devel@xxxxxxxxxxxxxxxxxxxx
> Cc: Daniel De Graaf <dgdegra@xxxxxxxxxxxxx>; Ian Jackson
> <Ian.Jackson@xxxxxxxxxx>
> Subject: Re: [PATCH v4 7/8] dm_op: convert HVMOP_inject_trap and
> HVMOP_inject_msi
> 
> On 17/01/17 17:29, Paul Durrant wrote:
> > diff --git a/xen/include/public/hvm/dm_op.h
> b/xen/include/public/hvm/dm_op.h
> > index 8e9bef1..abe8bbe 100644
> > --- a/xen/include/public/hvm/dm_op.h
> > +++ b/xen/include/public/hvm/dm_op.h
> > @@ -273,6 +273,52 @@ struct xen_dm_op_set_mem_type {
> >      uint64_aligned_t first_pfn;
> >  };
> >
> > +/*
> > + * XEN_DMOP_inject_trap: Inject a trap into a VCPU, which will get taken
> up
> > + *                       when it is next scheduled.
> > + *
> > + * Note that the caller should know enough of the state of the CPU before
> > + * injecting, to know what the effect of injecting the trap will be.
> > + */
> > +#define XEN_DMOP_inject_trap 13
> > +
> > +struct xen_dm_op_inject_trap {
> 
> Along with internal fixing of the (mis)use of the term trap inside the
> hypervisor, please s/trap/event/ in this new API.
> 

Ok.

> > +    /* IN - index of vCPU */
> > +    uint32_t vcpuid;
> > +    /* IN - interrupt vector */
> > +    uint8_t vector;
> > +    /* IN - trap type (DMOP_TRAP_* ) */
> > +    uint8_t type;
> > +/* NB. This enumeration precisely matches hvm.h:X86_EVENTTYPE_* */
> > +# define XEN_DMOP_TRAP_ext_int    0 /* external interrupt */
> > +# define XEN_DMOP_TRAP_nmi        2 /* nmi */
> > +# define XEN_DMOP_TRAP_hw_exc     3 /* hardware exception */
> > +# define XEN_DMOP_TRAP_sw_int     4 /* software interrupt (CD nn) */
> > +# define XEN_DMOP_TRAP_pri_sw_exc 5 /* ICEBP (F1) */
> > +# define XEN_DMOP_TRAP_sw_exc     6 /* INT3 (CC), INTO (CE) */
> > +    /* IN - enstruction length */
> 
> What is an enstruction? :)
> 

That would be an historic speeling mistake that I cut'n'pasted :-)

> Otherwise, Reviewed-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>

Ta,

  Paul


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