Greetings,
On 23.11.23 at 15:02, Bruce Ashfield wrote:
On Thu, Nov 23, 2023 at 3:51 AM Alexander Kanavin
<[email protected]> wrote:
On Wed, 22 Nov 2023 at 21:04, Michael Opdenacker via
lists.openembedded.org
<[email protected]> wrote:
The first prototype would only target the below architectures:
- "genericx86-64" architecture through the "genericx86-64" machine
(https://git.yoctoproject.org/poky/tree/meta-yocto-bsp/conf/machine/genericx86-64.conf)
- "arm64" architecture through the "qemuarm64" (and later "genericarm64"
when
hready) machine
(https://git.yoctoproject.org/poky/tree/meta/conf/machine/qemuarm64.conf)
Testing would happen through QEMU system emulation.
I have to question these choices, particularly generix86-64.
During the reviews, this was supposed to be qemux86-64 to match the
current qemuarm64 target. I'm not sure how genericx86-64 is in the
mailout .. probably just a lot of crossing email and review comments.
But regardless, these are just the machines for the prototype, which
won't be tested on hardware, but exclusively on qemu*.
Richard suggested the change, but I'd prefer to revert to qemux86-64 to
focus on the primary objectives.
After all, this prototype is meant to be used to test tooling and
upgrades between versions, and not to be used for testing on other
machines, right?
My understanding was that we're targeting specific goals, rather that
trying to be too open and too wide and lose the focus. At this stage,
are we interested in people testing binaries on other machines than
qemux86-64 and qemu-arm64? What matters is the tooling, processes and
policies that we develop, right?
Therefore, I don't see the point in having tunes that are different from
the ones used for the qemux86-64 and qemuarm64 machines.
So, here's what the scope looks like:
https://wiki.yoctoproject.org/wiki/Binary_Distro_Prototype#Target_architectures_and_machines
If you disagree, please try to convince me otherwise :)
Part of the prototype is to define a process as to how other 'official'
machines / architectures can be added (technical criteria, etc).
It is very important that the binary distribution makes a good first
impression at runtime, and a major part of that is that the common x86
hardware, including various peripherals, just works. Which is not the
case with genericx86-64: machines from meta-intel are a far better
choice.
Everything in the prototype needs to be on core, and not specific to
any vendor. i.e. handle them similarly to the reference boards
for oe-core/poky.
Which would mean something like the tunes from meta-intel could
be used if they are in core, but not linux-intel. That's part ot the criteria
that are being part of the outputs of the prototype. They aren't in this
initial email to keep it from being way too long to read.
Thanks for this input. I added it to a new section for drafting our
policies and processes:
https://wiki.yoctoproject.org/wiki/Binary_Distro_Prototype#Policies_and_Processes
Cheers
Michael.
--
Michael Opdenacker, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#1874):
https://lists.openembedded.org/g/openembedded-architecture/message/1874
Mute This Topic: https://lists.openembedded.org/mt/102755562/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-architecture/unsub
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
