[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH v4] x86/vm_event: Allow overwriting Xen's i-cache used for emulation



>>> On 22.09.16 at 20:54, <tamas.lengyel@xxxxxxxxxxxx> wrote:
> When emulating instructions Xen's emulator maintains a small i-cache fetched
> from the guest memory. This patch extends the vm_event interface to allow
> overwriting this i-cache via a buffer returned in the vm_event response.
> 
> When responding to a SOFTWARE_BREAKPOINT event (INT3) the monitor subscriber
> normally has to remove the INT3 from memory - singlestep - place back INT3
> to allow the guest to continue execution. This routine however is 
> susceptible
> to a race-condition on multi-vCPU guests. By allowing the subscriber to return
> the i-cache to be used for emulation it can side-step the problem by returning
> a clean buffer without the INT3 present.
> 
> As part of this patch we rename hvm_mem_access_emulate_one to
> hvm_emulate_one_vm_event to better reflect that it is used in various 
> vm_event
> scenarios now, not just in response to mem_access events.
> 
> Signed-off-by: Tamas K Lengyel <tamas.lengyel@xxxxxxxxxxxx>
> Acked-by: Razvan Cojocaru <rcojocaru@xxxxxxxxxxxxxxx>

Non-VM-event specific code:
Reviewed-by: Jan Beulich <jbeulich@xxxxxxxx>

One question though:

> --- a/xen/arch/x86/vm_event.c
> +++ b/xen/arch/x86/vm_event.c
> @@ -209,11 +209,20 @@ void vm_event_emulate_check(struct vcpu *v, 
> vm_event_response_t *rsp)
>          if ( p2m_mem_access_emulate_check(v, rsp) )
>          {
>              if ( rsp->flags & VM_EVENT_FLAG_SET_EMUL_READ_DATA )
> -                v->arch.vm_event->emul_read_data = rsp->data.emul_read_data;
> +                v->arch.vm_event->emul.read = rsp->data.emul.read;
>  
>              v->arch.vm_event->emulate_flags = rsp->flags;
>          }
>          break;
> +
> +    case VM_EVENT_REASON_SOFTWARE_BREAKPOINT:
> +        if ( rsp->flags & VM_EVENT_FLAG_SET_EMUL_INSN_DATA )
> +        {
> +            v->arch.vm_event->emul.insn = rsp->data.emul.insn;
> +            v->arch.vm_event->emulate_flags = rsp->flags;
> +        }
> +        break;

Is this intentionally different from the case above (where the setting
of ->emulate_flags is outside the inner if()?

Jan


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

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.