Re: [PATCH 1/5] dt-bindings: arm: altera: fix Agilex72 compatible vendor prefix
Rob Herring <[email protected]>
| Newsgroups | org.kernel.vger.linux-kernel,org.kernel.vger.linux-clk,org.kernel.vger.linux-devicetree |
|---|---|
| Message-ID | <[email protected]> |
On Tue, Aug 18, 2026 at 04:19:22PM +0800, Chiam, Evon wrote: > > > On 15/8/2026 1:35 am, Rob Herring wrote: > > On Fri, Aug 14, 2026 at 04:18:59PM +0800, Chiam, Evon wrote: > > > > > > > > > On 12/8/2026 10:49 am, Rob Herring wrote: > > > > On Thu, Aug 06, 2026 at 05:25:01PM +0800, [email protected] wrote: > > > > > From: "Chiam, Evon" <[email protected]> > > > > > > > > > > Altera has separated from Intel. Update the Agilex72 SoC and board > > > > > compatible strings to use the Altera vendor prefix instead of intel. > > > > > > > > Sorry, without better reasoning, you are stuck with them as sashiko > > > > pointed out. > > > > > > > Sashiko’s comment on the DTS still using the intel compatible strings was > > > based on reviewing patch 1/5 alone. Patch 4/5 moves the Agilex72 DT from > > > arch/arm64/boot/dts/intel/ to arch/arm64/boot/dts/altera/ and updates the > > > compatible strings to altr. With both patches applied, the bindings and DTS > > > remain consistent. > > > > > > Would you prefer that I combine the patches 1/5 and 4/5 into a single patch > > > so the binding and DTS updates land together? > > > > Moving the dts directory is fine (though that is really up to SoC > > maintainers). Changing compatible strings is not. It is not about the > > patch split. It is an ABI. There are exceptions though. Generally that > > is if there is no one with the h/w or the current .dts is so incomplete > > that anyone using the h/w will need to update their DTB anyways. > > > > Rob > > Thanks for clarifying. > > To provide some context, Agilex72 DTS is still at an early > stage of development, and most people do not have access to the > hardware yet. Anyone using the current DTS would need to update their > DTB, so changing the compatible strings at this stage should > not cause ABI issues. This information needs to go in the commit msg. Rob