Re: [PATCH v1 2/4] hvm/svm: Enable Breakpoint events

[Alexandru Stefan ISAILA <aisaila@bitdefender.com>]

On Vi, 2018-02-02 at 15:58 +0000, Andrew Cooper wrote:
> On 02/02/18 09:37, Alexandru Isaila wrote:
> >
> > This commit enables the breakpoint events for svm.
> >
> > Signed-off-by: Alexandru Isaila <aisaila@bitdefender.com>
> > ---
> >  xen/arch/x86/hvm/svm/svm.c    | 52
> > ++++++++++++++++++++++++++++++++++++-------
> >  xen/include/asm-x86/monitor.h |  3 ++-
> >  2 files changed, 46 insertions(+), 9 deletions(-)
> >
> > diff --git a/xen/arch/x86/hvm/svm/svm.c
> > b/xen/arch/x86/hvm/svm/svm.c
> > index dcbd550..14a5f60 100644
> > --- a/xen/arch/x86/hvm/svm/svm.c
> > +++ b/xen/arch/x86/hvm/svm/svm.c
> > @@ -59,6 +59,7 @@
> >  #include <asm/hap.h>
> >  #include <asm/apic.h>
> >  #include <asm/debugger.h>
> > +#include <asm/hvm/monitor.h>
> >  #include <asm/xstate.h>
> >
> >  void svm_asm_do_resume(void);
> > @@ -1079,7 +1080,8 @@ static void svm_ctxt_switch_to(struct vcpu
> > *v)
> >  static void noreturn svm_do_resume(struct vcpu *v)
> >  {
> >      struct vmcb_struct *vmcb = v->arch.hvm_svm.vmcb;
> > -    bool_t debug_state = v->domain->debugger_attached;
> > +    bool_t debug_state = v->domain->debugger_attached
> > +                || v->domain-
> > >arch.monitor.software_breakpoint_enabled;
> As a minor note, please clean up bool_t => bool as you end up making
> changes.
Will do, must of passed it.
>
> >
> >      bool_t vcpu_guestmode = 0;
> >      struct vlapic *vlapic = vcpu_vlapic(v);
> >
> > @@ -2407,6 +2409,23 @@ static bool svm_get_pending_event(struct
> > vcpu *v, struct x86_event *info)
> >      return true;
> >  }
> >
> > +static void svm_propagate_intr(struct vcpu *v, unsigned long
> > insn_len)
> > +{
> > +    struct vmcb_struct *vmcb = v->arch.hvm_svm.vmcb;
> > +    struct x86_event event = {
> > +        .vector = vmcb->eventinj.fields.type,
> > +        .type = vmcb->eventinj.fields.type,
> > +        .error_code = vmcb->exitinfo1,
> > +    };
> > +
> > +    if ( event.type >= X86_EVENTTYPE_SW_INTERRUPT )
> > +        event.insn_len = insn_len;
> > +    else
> > +        event.insn_len = 0;
> IIRC, you need to always set insn_len.  The length handling is vastly
> complicated (depends on event type and hardware availability, and in
> some cases needs emulating anyway), but the lower injection levels
> should DTRT.
>
> If they don't, can you provide a concrete example which doesn't work
> and
> we can see about what to do.
I've copied the functionality form vmx but I can remove the if
statement and always add the inst_len to the event.

Alex

________________________
This email was scanned by Bitdefender
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel