On 09/22/2014 10:32 AM, Richard Schmitt wrote:
Freescale will break every product layer out there for no good reason, it
seems. The
previous repository cannot be removed without critical impacts.
I don't see that. Currently meta-fsl-ppc is only used by customers building
for QorIQ based boards. So essentially we are renaming meta-fsl-ppc to be
meta-fsl-qoriq.
Then as we add our ARM support for LS1 and LS2, they will be BSPs under
meta-fsl-qoriq.
FSL processors are the ONLY one in the World to provide DPAA?
Yes, they are. and logically it is possible for both our ARM and PPC products
to support it. It is why we want a meta-fsl-qoriq layer, so that we could put
things like meta-fsl-dpaa in it.
We are trying to model the intel support. There is a meta-intel layer with
individual bsps and a common layer underneath it.
This seems to be the best approach to handle a collection of related products
out there.
I think it is too soon to talk about any reorganization of the
layers. I am still awaiting for the layers cleanup on the
meta-fsl-ppc as several recipes there belong to other layers (meta-networking,
meta-oe and so on).
That doesn't make sense to me. The idea is to cleanup the layers as part of
the restructuring. Why would we want to do it as separate steps.
When can we expect to see the restructuring take place?
Also, can we please get a ballpark estimate on when further
infrastructure support will come online as discussed in previous QorIQ
planning emails (e.g., Zhenhua's "The proposal of FSL QorIQ SDK
upstream" email on 7/16/14).
Specifically, I would like to get an idea on when I can expect to see
the following:
• Setup public bug management system for FSL SDKs
• Setup an external mailing list for external git repository for FSL SDK
discussion, patch submission, patch review.
Is this going to happen this year? I don't know whether I should depend
on this happening, but I view it as important to my project.
I have to believe that Freescale has internally numerous patches to
SDK1.6 that they don't make available to their customers unless a
service request is entered. This is a tremendous duplication of effort
and causes customer wheels to spin. I'm counting on soon seeing a
process in place that makes patches available to the 3.12 kernel that
shipped with SDK1.6. Am I being foolish?
thank you,
Bob
I have customers moving from/to TI and FSL, those use meta-ti and fsl-arm in
SAME
setup with no issues. It is as easy as change the machine name.
That may be fine for generic arm boards, but our BSPs will be extending the
generic architectures.
Rich
-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of
Otavio Salvador
Sent: Monday, September 22, 2014 9:15 AM
To: Bob Cochran
Cc: Luo Zhenhua-B19537; [email protected]; Schmitt
Richard-B43082; Brownfield Phil-RTJE20; Weng White-B18292
Subject: Re: [meta-freescale] The Yocto layer re-architect for FSL QorIQ SDK
On Mon, Sep 22, 2014 at 10:43 AM, Bob Cochran <[email protected]> wrote:
On 09/22/2014 08:51 AM, Otavio Salvador wrote:
On Mon, Sep 22, 2014 at 7:20 AM, [email protected]
<mailto:[email protected]> <[email protected]
<mailto:[email protected]>> wrote:
__
Following is the structure of the meta-fsl-qoriq layer. __ __
meta-fsl-qoriq____
|--- common # the common recipes for QorIQ ARM and
[snip]
__ __
Can you please review? Your comments and suggestions are welcome.
I think it is too soon to talk about any reorganization of the
layers. I am still awaiting for the layers cleanup on the
meta-fsl-ppc as several recipes there belong to other layers (meta-networking,
meta-oe and so on).
We also don't have the list of common recipes, so at this moment this
is still a guess about the common recipes.
LS1 is merged in fsl-arm so you can include it into FSL SDK in your
git submodules/supergit/repo file.
I see no benefit merging it, just cons ...
The benefit I see is that it logically groups all QorIQ parts together, which I
believe is important.
This does not make sense, sorry. If this would be the case you'd merge it in
OE-Core as you also use GCC, GLibC and so on.
When Freescale needs to provide support for LS2A (next generation) DPAA, the
current organization will become a mess. A lot (if not most) of the QorIQ BSP
layer is for DPAA / networking, and I don't think it will make sense to have
these duplicated in both meta-fsl-ppc and meta-fsl-arm, or have DPAA code that
will be built for the ARM in meta-fsl-ppc (or vice versa).
FSL processors are the ONLY one in the World to provide DPAA? I will provide
two possible routes:
- If FSL is the only DPAA provider, we make a mets-fsl-dpaa layer
- If NOT we move DPAA to meta-networking layer where it seems to belong
I would hope that having this code organized in logical layers under the scope
of QorIQ would make it more manageable in migrating from something like a TXXX
device to an LS2A device.
I have customers moving from/to TI and FSL, those use meta-ti and fsl-arm in
SAME setup with no issues. It is as easy as change the machine name.
How users from meta-fsl-ppc will migrated for any new layer layout?
Freescale will break every product layer out there for no good reason, it
seems. The previous repository cannot be removed without critical impacts.
...
--
_______________________________________________
meta-freescale mailing list
[email protected]
https://lists.yoctoproject.org/listinfo/meta-freescale