[meta-intel] [dizzy][PATCH 2/3] meta-crystalforest: add qat preferred provider

Mittal, AnujX anujx.mittal at intel.com
Thu Jul 23 16:39:25 PDT 2015


> -----Original Message-----
> From: Darren Hart [mailto:dvhart at linux.intel.com]
> Sent: Friday, July 24, 2015 5:44 AM
> To: Mittal, AnujX; meta-intel at yoctoproject.org; Ong, Boon Leong
> Subject: Re: [meta-intel] [dizzy][PATCH 2/3] meta-crystalforest: add qat preferred provider
> 
> 
> 
> On 7/23/15 9:38 AM, Anuj Mittal wrote:
> > Added the qat preferred provider to be qat16.
> >
> > Signed-off-by: Anuj Mittal <anujx.mittal at intel.com>
> > ---
> >  meta-crystalforest/conf/machine/crystalforest.conf |    2 ++
> >  1 file changed, 2 insertions(+)
> >
> > diff --git a/meta-crystalforest/conf/machine/crystalforest.conf b/meta-
> crystalforest/conf/machine/crystalforest.conf
> > index 44b9bb2..84a6df2 100644
> > --- a/meta-crystalforest/conf/machine/crystalforest.conf
> > +++ b/meta-crystalforest/conf/machine/crystalforest.conf
> > @@ -18,6 +18,8 @@
> >  PREFERRED_PROVIDER_virtual/kernel ?= "linux-yocto"
> >  PREFERRED_VERSION_linux-yocto ?= "3.10%"
> >
> > +PREFERRED_PROVIDER_virtual/qat ?= "qat16"
> 
> Your cover letter said crystal forest depends on qat15. Why are we
> setting it to qat16 here?

Default is going to remain qat16 for highland forest platforms. crystal forest & highland forest platforms use the same machine - crystalforest.

> 
> Is there a reason we are doing this as qat15 and qat16 instead of just
> having two qat recipes with different versions and setting the
> PREFERRED_VERSION_qat in the machine config?
> 
> e.g.
> 
> qat_15-1.7.0-30.bb
> qat_16-2.2.0-30.bb
> 
> PREFERRED_VERSION_qat ?= "15-%"
> 
> There shouldn't be a need for a virtual/qat at all.

Both qat15 & qat16 come from the same package. So PV remains the same between these two.

I think it was decided to keep separate names like this to avoid any problems with the upgrade path because of same PV.

> 
> --
> Darren Hart
> Intel Open Source Technology Center


More information about the meta-intel mailing list