Message ID | 20240125151630.753318-1-tudor.ambarus@linaro.org (mailing list archive) |
---|---|
Headers | show |
Series | ARM: dts: samsung: specify the SPI fifosize | expand |
On Thu, Jan 25, 2024 at 03:16:23PM +0000, Tudor Ambarus wrote: > These patches close the circle and break the dependency between the SPI > of_alias ID and the SPI driver. The SPI of_alias ID was used as an index > into the fifo_lvl_mask to determine the FIFO depth of the SPI node. > Changing the alias ID into the device tree would make the driver choose > a wrong FIFO size configuration, if not accessing past the fifo_lvl_mask > array boundaries. Not specifying an SPI alias would make the driver fail > to probe, which was wrong too. Thus I updated the driver and I provided Specifying a specific compatible is just good practice, it makes the DT much more robust for when we discover new things about the hardware and reduces the surface of the ABI.
On 25/01/2024 16:16, Tudor Ambarus wrote: > Hi, > > The "samsung,spi-fifosize" was introduced in the following patch set: > https://lore.kernel.org/linux-arm-kernel/20240125145007.748295-1-tudor.ambarus@linaro.org/T/#t > > I expect Mark will provide an immutable tag for the bindings so that it > can be merged into the samsung tree. We'll avoid this way using the > property into the device trees without having it defined into the > bindings. No need. Any merge back to DTS branch would anyway question my pull request to arm soc, so I am not going to do this, unless absolutely necessary. I'll take the DTS once binding is merged to ASoC, so everything will be fine as long linux-next is fine. Best regards, Krzysztof
On 25/01/2024 16:16, Tudor Ambarus wrote: > Hi, > > The "samsung,spi-fifosize" was introduced in the following patch set: > https://lore.kernel.org/linux-arm-kernel/20240125145007.748295-1-tudor.ambarus@linaro.org/T/#t > Judging by Rob's statement the bindings were not accepted, so I drop this series for now. Best regards, Krzysztof