[meta-intel] [fido][PATCH 0/3] QAT recipe for Intel Riverforest

Ong, Boon Leong boon.leong.ong at intel.com
Mon Jul 6 15:55:07 PDT 2015



>-----Original Message-----
>From: Saul Wold [mailto:sgw at linux.intel.com]
>Sent: Monday, July 6, 2015 11:38 PM
>To: Mittal, AnujX; meta-intel at yoctoproject.org
>Cc: Ong, Boon Leong
>Subject: Re: [meta-intel] [fido][PATCH 0/3] QAT recipe for Intel Riverforest
>
>On 07/05/2015 10:28 PM, anujx.mittal at intel.com wrote:
>> From: Anuj Mittal <anujx.mittal at intel.com>
>>
>> Dear Maintainer(s),
>>
>> These patches provide the support for Intel QAT v 2.3.0-34 on
>> RiverForest platforms.
>>
>Is the intent also of this patch set to allow for an upgrade from Dizzy's QAT16
>(2.2.x based) -> this version?  Or is there no upgrade path on the device for
>something already installed?

Anuj's focus is on this QAT v2.3.0-34 version for RiverForest platform on fido.
Venkat's work on Crystal Forest and Highland Forest were on dizzy on older QAT version. 
There is no plan for now to upgrade the Highland Forest and Crystal Forest device support
on Fido due to resource constraints and no strong marketing ask.

Having said above, it is important to have at least matching filename, structure and reuse
as much as common ingredient to minimize divergence in meta-data which Saul rightful
pointed that out below. 

>
>If it's intended as an upgrade then renaming from qat16 -> qat will allow for that
>to happen with out some additional METADATA, such as the PROVIDES_<pkg> /
>RPROVIDES_<pkg> and RREPLACES_<pkg>.
>
>
>Are you coordinating with venkatasubramanianx.pattu at intel.com, who is working
>with the Dizzy based patches?  It seems that these should be sharing a similar
>base.  Maybe work together to review changes the changes.
Yes, common code base is important. 

>
>The Dizzy based QAT recipes are using the correct $B and $S correctly, please
>consider looking at that.
>
>I attempted to build your version here and it did not even unpack the
>QAT1.6/QAT1.6.L.2.3.0-34.tar.gz.
>
>The do_install_prepend should not be doing a blind "cp -Pr ${B}/* ${D}".
>


More information about the meta-intel mailing list