On 2/21/24 9:06 AM, Paul Barker wrote:
On 21/02/2024 10:57, Ross Burton wrote:
From: Ross Burton <[email protected]>
This is a new 64-bit "generic" Arm machine, that expects the hardware to
be SystemReady IR compatible. This is slightly forward-leaning as there's
not a _lot_ of SystemReady hardware in the wild, but most modern boards
are and the number will only grow. Also, this is the only way to have a
'generic' machine as without standardised bootloaders and firmware it
would be impossible.
The base machine configuration isn't that exciting: it's a fully featured
machine that supports most things, booting via UEFI and an initramfs.
However, the kernel is more interesting. This RFC uses the upstream defconfig
because unlike some other platforms, the arm64 defconfig is actively
maintained with the goal of being a 'boots on most hardware' configuration.
My argument is: why would we duplicate that effort?
The "linux-yocto way" is configuration fragments and after a week of
hair-pulling I do actually have fragments that boot on a BeaglePlay, but
to say this was a tiresome and frustrating exercise would be understating it.
So, a request for comments: is it acceptable to use the upstream defconfig in
a reference BSP? Personally I'm torn: the Yocto way is fragments not monolithic
configs, but repeating the effort to fragmentise the configuration and then
also have it sufficiently modular that it can be used in pieces - instead of
just being a large file split up into smaller files - is a lot of effort for
what might end up being minimal gain. My fear is we end up with a fragmented
configuration that can't be easily modified without breaking some platforms,
and badly copies what the defconfig already does.
I am in favour of this - I think the "genericarm64" machine should use
the in-tree defconfig so that it can support the widest array of
hardware. If someone wants to trim down the kernel for a particular
platform then they should probably create a specific MACHINE anyway.
If we take the other approach of building up the kernel config from
fragments, how would we know that all SystemReady IR capable systems
will be supported? Yocto Project doesn't have the resources to test
every platform.
I disagree here. I think it would be MUCH better to have a 'SystemReady IR'
hardware configuration. So if SystemReady IR is desired, it is something that
anyone can enable (starting with genericarm64). Remember the defconfig is going
to have more then hardware configs in it. Will it have the right systemd
configurations? Will is have the magic filesystem a random user wants? Will
avoid having some other filesystem type that another user doesn't want?
Building up the kernel, and considering SystemReady IR as a 'hardware feature',
and then add in the additional things that are needed for whatever reason is a
much more reasonable way to do this and make it useful to otthers.
For the Renesas RZ SoCs I work on these days, the in-tree defconfig is
the configuration we test with the mainline kernel.
AMD does the same thing, for the kernel development it makes sense. Kernel is
built and tested standalone from userspace.
But with that said, I think it's the wrong way to do Yocto Project development.
Yocto Project development needs further control and the separation of hardware
and software configurations is pretty essential to having a system that can be
customized appropriately.
The defconfig can be used as a guide to the other configurations, but separating
hardware and software configs is a necessary first step in my opinion.
--Mark
Thanks,
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#196005):
https://lists.openembedded.org/g/openembedded-core/message/196005
Mute This Topic: https://lists.openembedded.org/mt/104502773/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
