Re: [PATCH 15 of 16] libxl: Introduce a new guest config file parameter [and 1 more messages]

[Wei Wang2 <wei.wang2@amd.com>]

On Tuesday 10 January 2012 18:12:05 Ian Jackson wrote:
> Wei Wang writes ("[PATCH 14 of 14 V3] libxl: Introduce a new guest config 
file parameter"):
> > # HG changeset patch
> > # User Wei Wang <wei.wang2@amd.com>
> > # Date 1326213623 -3600
> > # Node ID 39eb093ea89eeaa4dbff29439499f2a289291ff0
> > # Parent  9e89b6485b6c91a8d563c46c47a8d768eee7d1f2
> > libxl: Introduce a new guest config file parameter
> > Use iommu = {1,0} to enable or disable guest iommu emulation.
> > Default value is 0.
>
> Ian Jackson writes ("Re: [PATCH 15 of 16] libxl: Introduce a new guest 
config file parameter"):
> > Wei Wang writes ("[PATCH 15 of 16] libxl: Introduce a new guest config 
file parameter"):
> > > libxl: Introduce a new guest config file parameter
> > > Use iommu = {1,0} to enable or disable guest iommu emulation.
> > > Default value is 0.
> >
> > I'm not sure I like this name.  It's confusing because it's not at
> > first glance clear whether it refers to the host's putative iommu, or
> > some kind of provision to the guest.
How about guest_iommu = {1, 0} or ats_passthru = {1, 0}?  Actually the idea of 
the whole patch queue is to enable passthru  for sophisticated ats devices 
(e.g. Tahiti the latest AMD gfx/gpgpu), so that we utilize the device to do 
general-purpose computations in guest. 

> > And it needs documentation, which I hope will answer my other question
> > which is "what is this good for?" :-).

Sure, I am working on the docs now. It will be sent together with the next try 
after I had collected enough comments of this version.

Thanks,
Wei
> Ian.