On Mon, 4 Mar 2024, Conor Dooley wrote: > On Mon, Mar 04, 2024 at 11:59:47AM +0100, Sebastian Andrzej Siewior wrote: > > | $ git grep fu540-c000-uart > > | Documentation/devicetree/bindings/serial/sifive-serial.yaml: - sifive,fu540-c000-uart > > | Documentation/devicetree/bindings/serial/sifive-serial.yaml: compatible = "sifive,fu540-c000-uart", "sifive,uart0"; > > | Documentation/devicetree/bindings/sifive/sifive-blocks-ip-versioning.txt:"sifive,fu540-c000-uart". This way, if SoC-specific > > | Documentation/devicetree/bindings/sifive/sifive-blocks-ip-versioning.txt: compatible = "sifive,fu540-c000-uart", "sifive,uart0"; > > | arch/riscv/boot/dts/sifive/fu540-c000.dtsi: compatible = "sifive,fu540-c000-uart", "sifive,uart0"; > > | arch/riscv/boot/dts/sifive/fu540-c000.dtsi: compatible = "sifive,fu540-c000-uart", "sifive,uart0"; > > | drivers/tty/serial/sifive.c:OF_EARLYCON_DECLARE(sifive, "sifive,fu540-c000-uart0", > > | drivers/tty/serial/sifive.c: { .compatible = "sifive,fu540-c000-uart0" }, > > > > note that the driver has a trailing 0 in the binding while the yaml > > description and the DT part does not. > > The 'sifive,uart' has a trailing 0 where the 0 denotes the version UART > > IP. > > > > Was this also intended for the fu540-c000-uart binding? Should the 0 be > > added everywhere or removed from the driver? > > I suspect that the driver is what's incorrect, given there's little > value in putting the IP version in the SoC-specific compatible as it's > a fixed implementation. I'd change the driver to match the bindings. Agreed - Paul