On Mon Feb 23, 2026 at 4:43 PM CST, Andrew Davis wrote:
> On 2/23/26 3:54 PM, Ryan Eatmon wrote:
>> 
>> 
>> On 2/23/2026 3:51 PM, Andrew Davis wrote:
>>> On 2/23/26 2:22 PM, Randolph Sapp via lists.yoctoproject.org wrote:
>>>> From: Randolph Sapp <[email protected]>
>>>>
>>>> Add a TI configuration for the BeagleY-AI development board.
>>>>
>>>> Signed-off-by: Randolph Sapp <[email protected]>
>>>> ---
>>>>   meta-ti-bsp/conf/machine/beagley-ai-ti-k3r5.conf |  7 +++++++
>>>>   meta-ti-bsp/conf/machine/beagley-ai-ti.conf      | 16 ++++++++++++++++
>>>>   2 files changed, 23 insertions(+)
>>>>   create mode 100644 meta-ti-bsp/conf/machine/beagley-ai-ti-k3r5.conf
>>>>   create mode 100644 meta-ti-bsp/conf/machine/beagley-ai-ti.conf
>>>>
>>>> diff --git a/meta-ti-bsp/conf/machine/beagley-ai-ti-k3r5.conf 
>>>> b/meta-ti-bsp/conf/machine/beagley-ai-ti-k3r5.conf
>>>> new file mode 100644
>>>> index 00000000..88d0888b
>>>> --- /dev/null
>>>> +++ b/meta-ti-bsp/conf/machine/beagley-ai-ti-k3r5.conf
>>>> @@ -0,0 +1,7 @@
>>>> +#@TYPE: Machine
>>>> +#@NAME: BeagleY-AI (R5F)
>>>> +#@DESCRIPTION: Machine configuration for the BeagleY-AI (R5F core)
>>>> +
>>>> +require conf/machine/include/k3r5.inc
>>>> +
>>>> +UBOOT_MACHINE = "am67a_beagley_ai_r5_defconfig"
>>>> diff --git a/meta-ti-bsp/conf/machine/beagley-ai-ti.conf 
>>>> b/meta-ti-bsp/conf/machine/beagley-ai-ti.conf
>>>> new file mode 100644
>>>> index 00000000..088cbd62
>>>> --- /dev/null
>>>> +++ b/meta-ti-bsp/conf/machine/beagley-ai-ti.conf
>>>> @@ -0,0 +1,16 @@
>>>> +#@TYPE: Machine
>>>> +#@NAME: BeagleY-AI (A53)
>>>> +#@DESCRIPTION: Machine configuration for the BeagleY-AI (A53)
>>>> +
>>>> +require conf/machine/include/j722s.inc
>>>> +
>>>> +KERNEL_DEVICETREE_PREFIX = " \
>>>> +    ti/k3-am67a \
>>>> +    ti/k3-j722s \
>>>> +"
>>>> +
>>>> +KERNEL_DEVICETREE = " \
>>>> +    ti/k3-am67a-beagley-ai.dtb \
>>>> +"
>>>> +
>>>> +UBOOT_MACHINE = "am67a_beagley_ai_a53_defconfig"
>>>
>>> This defconfig doesn't work if you select an older BSP.
>>>
>>> Thinking on this, the only difference we should have between this machine
>>> config and the one already in meta-beagle is the default selected BSP
>>> (bsp-ti-6_12 vs bb_org-6_12). Why can't we just have the one config and
>>> select the BSP with TI_PREFERRED_BSP? We could do that externally
>>> from the build env, or with a branding.
>>>
>>> The issue with this patch is we would now have two configs for the same
>>> hardware, and there is no TI produced BeagleY, so having the machine
>>> config for it in this layer just seems wrong. I have the same complaint
>>> for beagleplay-ti and beaglebadge-ti, we should drop those too and fix
>>> them in the same way.
>> 
>> The difference between the two platforms is that one is supported by TI and 
>> one is not.  We do not answer questions or support the meta-beagle boards.  
>> Those are supported by the community (aka Denys).  But the beagleplay-ti 
>> board using the TI kernel and TI uboot is supported.  We will answer 
>> questions about them.
>> 
>
> And selecting TI kernel and TI uboot is done by simply setting
> TI_PREFERRED_BSP = "ti-6_12", why do we need to clone files for this?
>
>> It was a decision from Sitara management to do this.
>> 
>
> To do what? Support BeagleY, sure that is fine, we also support Wayland
> for instance, but we don't go and clone it and rebrand it Wayland-TI™.
>
> We support community projects by working in those communities, in this
> case the community project is "meta-beagle". So let's add support for
> our TI BSP to that project, not fork bits of it into our meta-ti project.
>
> Anyway to close the loop on the two threads,
>
>> I don't necessarily hate that idea. Guessing this is hinging on Andrew's 
>> idea of
>> overriding the BSP provider variable directly for our internal testing?
>
> We could at very least start with this "beagley-ai-ti.conf" do nothing but
> the following two lines:
>
> require meta-beagle/conf/machine/beagley-ai.conf
> TI_PREFERRED_BSP = "ti-6_18"
>
> This would remove any duplication at least, but does introduce a reverse
> dependency from meta-ti towards meta-beagle. So the first option of a
> BSP provider override as needed is still my strong preference.
>
> Andrew
>

I'd almost prefer we move all the beagle-ti configs back into meta-beagle and do
that override there. Makes the layer dependency tree more reasonable and
provides logical isolation from meta-ti and meta-beagle devices.

I'll draft that. Doesn't seem so awful.

>> That's why we name them differently to draw a distinction between the two.
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#19577): 
https://lists.yoctoproject.org/g/meta-ti/message/19577
Mute This Topic: https://lists.yoctoproject.org/mt/117964418/21656
Group Owner: [email protected]
Unsubscribe: https://lists.yoctoproject.org/g/meta-ti/unsub 
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to