On Thu, Jan 25, 2024 at 07:01:10PM +0000, Tudor Ambarus wrote: > > > On 1/25/24 17:45, Mark Brown wrote: > > On Thu, Jan 25, 2024 at 05:30:53PM +0000, Tudor Ambarus wrote: > >> On 1/25/24 17:26, Mark Brown wrote: > > > >>> OK, so just the compatible is enough information then? > > > >> For gs101, yes. All the gs101 SPI instances are configured with 64 bytes > >> FIFO depths. So instead of specifying the FIFO depth for each SPI node, > >> we can infer the FIFO depth from the compatible. > > > > But this is needed for other SoCs? This change is scattered through a > > There are SoCs that have multiple instances of the SPI IP, and they > configure them with different FIFO depths. See > "samsung,exynosautov9-spi" for example: SPI0, SPI1, and SPI6 are > configured by the SoC to use 256 bytes FIFO depths, while all the other > 8 SPI instances use 64 bytes FIFOs. I tried to explain the entire logic > of the series in another reply, see it here: > https://lore.kernel.org/linux-arm-kernel/40ba9481-4aea-4a72-87bd-c2db319be069@xxxxxxxxxx/T/#u We have some common properties for fifo size. In fact, there was just a discussion recently on Samsung UART (Is that the same block?) about this. So if you do use a property here, use a common one. Rob