On 12/01/2024 18:42, Patrick Williams wrote: > On Fri, Jan 12, 2024 at 06:14:26PM +0100, Krzysztof Kozlowski wrote: >> On 12/01/2024 18:10, Patrick Williams wrote: >>> On Fri, Jan 12, 2024 at 08:10:25AM +0100, Krzysztof Kozlowski wrote: >>>> On 12/01/2024 02:36, Delphine CC Chiu wrote: >>>>> Document the new compatibles used on Facebook Yosemite 4. >>>> >>>> There is Yosemite4 board already supported. What is this for? >>>> https://lore.kernel.org/all/20240109072053.3980855-5-Delphine_CC_Chiu@xxxxxxxxxx/ >>> >>> Yosemite4 is a server chassis which is managed by a BMC. The BMC is on >>> a pluggable module card. Typically we've used Aspeed chips for this, >>> but we are building an alternative BMC module using Nuvoton BMC chips. >> >> There are few ways to solve this, like having different compatibles or >> having some shared compatibles to note common part of hardware. However >> usually the final compatible represents the final device, which here you >> use for two entirely different products. This works only for the cases >> of carrier boards, where that compatible indeed represents the same >> hardware. >> >> Not your case. This needs fixing. > > This patch: > + model = "Facebook Yosemite 4 BMC"; > + compatible = "facebook,yosemite4-n-bmc", "nuvoton,npcm845"; > > Aspeed patch: > + model = "Facebook Yosemite 4 BMC"; > + compatible = "facebook,yosemite4-bmc", "aspeed,ast2600"; > > These have different compatibles already ('-n' for Nuvoton). Do we just > need the model to be clearly different also? Maybe there is something > else I'm not understanding. Ah, no, it's fine. It is just a bit confusing. However commit msg for sure misses a lot of this explanation. Please write something useful to avoid such discussions... Best regards, Krzysztof
