On 19.12.2024 09:11:58, Elaine Zhang wrote: > Is new controller: > Support CAN and CANFD protocol. > > Signed-off-by: Elaine Zhang <zhangqing@xxxxxxxxxxxxxx> Thanks for porting the rk3576 to the mainline driver. Looking at subtle differences between the IP cores I'm not sure if you want to add it to the existing driver or have a dedicated driver. But we can decide on this later. At least the mainline driver has a better structure than your original downstream driver. Why was the timestamp register removed from the registers? Previously the driver supported hardware timestamping. :( Some general comments here, a more detailed review will follow. Why have bits been removed from several registers and the remaining ones moved? That doesn't make it easier for driver developers to cover new IP cores. :( Please use FIELD_GET(), FIELD_PREP() macros instead of open coding shift and mask operations. See rest of the driver. Have you actually tested this code in both RX and TX direction? I don't see rkcanfd_handle_tx_int() being called? What's the purpose of "rockchip,rx-max-data"? Why do you add "rx_fifo_depth" to "struct rkcanfd_priv"? It's not used outside of rkcanfd_probe(). Can you configure the IP core for CAN-2.0 only mode, so that it only receives CAN-2.0 frames only? regards, Marc -- Pengutronix e.K. | Marc Kleine-Budde | Embedded Linux | https://www.pengutronix.de | Vertretung Nürnberg | Phone: +49-5121-206917-129 | Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-9 |
Attachment:
signature.asc
Description: PGP signature
