Re: [Xen-devel] [PATCH 2/6] vTPM: add HVM_PARAM_STUBDOM_VTPM parameter for HVM virtual machine

[Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>]

On Thu, 30 Oct 2014, Xu, Quan wrote:
> > -----Original Message-----
> > From: Stefano Stabellini [mailto:stefano.stabellini@xxxxxxxxxxxxx]
> > Sent: Thursday, October 30, 2014 9:35 PM
> > To: Andrew Cooper
> > Cc: Xu, Quan; xen-devel@xxxxxxxxxxxxx; ian.jackson@xxxxxxxxxxxxx; 
> > tim@xxxxxxx;
> > keir@xxxxxxx; ian.campbell@xxxxxxxxxx; jbeulich@xxxxxxxx
> > Subject: Re: [Xen-devel] [PATCH 2/6] vTPM: add
> > HVM_PARAM_STUBDOM_VTPM parameter for HVM virtual machine
> > 
> > On Thu, 30 Oct 2014, Andrew Cooper wrote:
> > > On 30/10/14 12:05, Xu, Quan wrote:
> > > >
> > > >> -----Original Message-----
> > > >> From: Andrew Cooper [mailto:andrew.cooper3@xxxxxxxxxx]
> > > >> Sent: Thursday, October 30, 2014 7:50 PM
> > > >> To: Xu, Quan; xen-devel@xxxxxxxxxxxxx
> > > >> Cc: keir@xxxxxxx; ian.campbell@xxxxxxxxxx; tim@xxxxxxx;
> > > >> ian.jackson@xxxxxxxxxxxxx; jbeulich@xxxxxxxx
> > > >> Subject: Re: [Xen-devel] [PATCH 2/6] vTPM: add
> > > >> HVM_PARAM_STUBDOM_VTPM parameter for HVM virtual machine
> > > >>
> > > >> On 30/10/14 07:38, Quan Xu wrote:
> > > >>> Signed-off-by: Quan Xu <quan.xu@xxxxxxxxx>
> > > >> What is the purpose of this parameter?  A patch like this is
> > > >> currently unacceptable, especially as the libxl hunk indicates that
> > > >> the parameter name does not match whatever information you are putting
> > into it.
> > > >>
> > > > Thanks for your suggestion.
> > > > This parameter tell the Qemu whether to register Qemu vTPM frontend in
> > xen_hvm_init().
> > > > Qemu will get the parameter value by xc_get_hvm_param(). How can I
> > change it?
> > >
> > > This is surely something which should be a command line parameter to
> > > qemu, or perhaps for qemu to read out of xenstore.
> > >
> > > An HVM param is entirely inappropriate for this purpose, in my opinion.
> > 
> > I agree that an HVM param for this might not the best way to do it, but I 
> > can see
> > why Quan did it that way as we already have a few key parameters passed to
> > QEMU that way.
> > 
> > A QEMU command line option, QMP command or xenstore key would be better.
> 
> If hvm param is not the best way, I think xenstore key would be better.
> 
> Below is part of Qemu patch, that's why I add HVM_PARAM_STUBDOM_VTPM param.
> xen_vtpm_register() is similar to xen_be_register()
> 
> ### Qemu : xen_hvm_init() [xen-hvm.c]###
> +#ifdef CONFIG_TPM_XENSTUBDOMS
> +    xc_get_hvm_param(xen_xc, xen_domid, HVM_PARAM_STUBDOM_VTPM, 
> &stubdom_vtpm);
> +    if (stubdom_vtpm) {
> +        xen_vtpm_register(&xen_vtpmdev_ops);
> +    }
> +#endif

I think I would need to see the rest of the QEMU patches to be able to
tell you which way I think is best.
In this context is vtpm an emulated device or a PV backend?

>     xen_be_register("console", &xen_console_ops);
>     xen_be_register("vkbd", &xen_kbdmouse_ops);
>     xen_be_register("qdisk", &xen_blkdev_ops);
>     xen_read_physmap(state);
> ##### Qemu ####
> 
> 
> Quan 
> 

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