On 28/03/2024 03:32, Pavan Holla wrote: > On Tue, Mar 26, 2024 at 9:59 PM Krzysztof Kozlowski <krzk@xxxxxxxxxx> wrote: >> >> On 27/03/2024 04:39, Pavan Holla wrote: >>> Hi Krzysztof, >>> >>> Thanks for the review. >>> >>> On Tue, Mar 26, 2024 at 1:47 AM Krzysztof Kozlowski <krzk@xxxxxxxxxx> wrote: >>>> Nothing improved. >>> >>> Yes. I only added maintainers of drivers/platform/chrome in v2. I am >>> still investigating why MODULE_ALIAS() is required. >> >> Heh, I wrote why. You miss ID table. > > This driver is going to be used by the cros_ec_dev.c MFD. The UCSI device doesn’t > have an ACPI or OF entry, so I am not sure how I can use MODULE_DEVICE_TABLE > here. If I don’t use MODULE_ALIAS(“platform:” DRV_NAME), > https://elixir.bootlin.com/linux/latest/source/drivers/mfd/cros_ec_dev.c#L206 > isn’t able to automatically associate the driver with the device at boot. > I haven’t upstreamed the change in cros_ec_dev.c yet, but the code is similar to > existing code for drivers/platform/chrome/cros_usbpd_logger.c. There are many > other occurrences of the same MODULE_ALIAS pattern: Just open other platform drivers and look how it is done there. Or ask colleagues. There is absolutely no one in entire Chromium/google who ever wrote platform_driver? platform_driver has ID table for matching. Otherwise how do you expect this to be matched? How your driver is being matched and device bound? By fallback, right? So what is the primary method? Best regards, Krzysztof
