Re: [PATCH v4 13/14] arm/libxl: Emulated PCI device tree node in libxl

[Ian Jackson <iwj@xxxxxxxxxxxxxx>]

Rahul Singh writes ("Re: [PATCH v4 13/14] arm/libxl: Emulated PCI device tree 
node in libxl"):
> Hi Ian        
> > What is wrong with putting it in
> > libxl__arch_domain_build_info_setdefault
> > which I think exists precisely for this kind of thing ?
> 
> As we have to set the arch_arm.vpci to false for x86 and ARM I
> thought it is right to move the code to common code to avoid
> duplication.
> 
> Are you suggesting to put "
> libxl_defbool_setdefault(&b_info->arch_arm.vpci, false)”in
> libxl__arch_domain_build_info_setdefault() for x86 and ARM
> differently.

I've gone back and reread the whole thread, which I probably should
have done to start with....

So:

> >> #if defined(__arm__) || defined(__aarch64__)                  
> >>    /*                                    
> >>     * Enable VPCI support for ARM. VPCI support for DOMU guests is not    
> >>     * supported for x86.                          
> >>     */                                   
> >>    if (d_config->num_pcidevs)                        
> >>      libxl_defbool_set(&b_info->arch_arm.vpci, true);           
> >> #endif 

I think this logic probably ought to be in libxl, not in xl.  We try
to make the libxl API "do the right thing" by default.  In this case I
think that means to enable VPCI (i) on platforms where it's available
(ii) if the guest has PCI passthrough devices.  Is that right ?

Sorry to ask these question now, and please forgive my ignorance:

Is VPCI inherently an ARM-specific ABI or protocol ?  When might an
admin want to turn it on explicitly ?

How does this all relate to the (non-arch-specific) "passthrough"
option ?

Ian.