[linux-yocto] [PATCH 1/1] [KERNEL] meta: Crystal Forest Machine Created.

[Darren Hart]

On 08/08/2012 11:32 AM, Bruce Ashfield wrote:
> On 12-08-08 02:11 PM, Darren Hart wrote:
>>
>>
>> On 08/08/2012 10:56 AM, kishore.k.bodke at intel.com wrote:
>>> From: Kishore Bodke<kishore.k.bodke at intel.com>
>>>
>>
>>> +++ b/meta/cfg/kernel-cache/bsp/crystalforest/crystalforest.scc
>>> @@ -0,0 +1,21 @@
>>> +kconf hardware crystalforest.cfg
>>> +
>>> +include cfg/x86_64.scc
>>> +include cfg/8250.scc
>>> +
>>> +include features/power/intel.scc
>>> +
>>> +#These are required features for Intel DPDK Support
>>> +include features/uio/uio.scc
>>> +include features/hugetlb/hugetlb.scc
>>> +include features/ixgbe/ixgbe.scc
>>> +include features/igb/igb.scc
>>> +
>>> +
>>> +include features/latencytop/latencytop.scc
>>> +include features/profiling/profiling.scc
>>> +include features/usb/xhci-hcd.scc
>>> +include features/usb/ehci-hcd.scc
>>> +include features/usb/ohci-hcd.scc
>>> +include cfg/usb-mass-storage.scc
>>> +include cfg/boot-live.scc
>>>
>>
>> These last two should be in the standard and preempt-rt scc files and
>> not in the base crystalforest.scc. The reason being, the kernel type
>> should define policy - and whether or not we support mass storage and
>> live boot is a matter of policy, not hardware enablement. The tiny
>> kernel should not be required to pull in usb mass storage, fat support,
>> etc to boot this machine.
> 
> (moving to linux-yocto).
> 
> Agreed on this, that the policy should be out of the BSPs.
> 
> There are other fragments to move around as well, which I'm currently
> looking at.
> 
> Here's the question, do we want this for all boards, all arches, or do
> we want some sort of intermediate x86/intel feature that defines this
> policy, and the BSPs include that ?
> 
> Either way works, and I'm inclined to move this to standard/rt as well,
> but I wanted to gather some other opinions as well.

For the purposes of Kishore's patches, I'm just looking for it to be
consistent with what's in the tree now.

For the long term, we shouldn't have to duplicate usb-mass-storage and
boot-live for every x86* BSP for both preempt-rt and standard. However,
we don't want that for the PPC, ARM, and MIPS machines, as far as I
understand it, as we don't have live images (hddimg) for them anyway.

Whether an hddimg is a valid image type or not is machine specific, so
I'm not sure there is a better (and still correct) way to describe it
than what we do currently (but I'd like to be wrong).

-- 
Darren Hart
Intel Open Source Technology Center
Yocto Project - Technical Lead - Linux Kernel