On Thu, May 6, 2021 at 4:37 PM Guenter Roeck <linux@xxxxxxxxxxxx> wrote: > On Thu, May 06, 2021 at 12:28:40PM +0300, Andy Shevchenko wrote: > > On Thu, May 6, 2021 at 6:43 AM Guenter Roeck <linux@xxxxxxxxxxxx> wrote: > > > > > > With CONFIG_ACPI=n and -Werror, 0-day reports: > > > > > > drivers/iio/chemical/bme680_i2c.c:46:36: error: > > > 'bme680_acpi_match' defined but not used > > > > > > Apparently BME0680 is not a valid ACPI ID. Remove it and with it > > > ACPI support from the bme680_i2c driver. > > > > Reviewed-by: Andy Shevchenko <andy.shevchenko@xxxxxxxxx> > > > > with the SPI part amended in the same way. > > > Right. I just sent a patch doing that. Oddly enough 0-day didn't complain > about that one to me, nor about many other drivers with the same problem. > No idea how it decides if and when to make noise. randconfig I believe. > Is there a way to determine invalid ACPI IDs ? I could write a coccinelle > script to remove the code automatically. As Hans said... My understanding that most of the fake IDs come into life due to: - people apply similar rules to them as they knew about OF case (and certain maintainers blindly allowed that) - people in big companies need to quickly prototype something without giving a crap about ACPI specification and / or process The last part (I believe the smallest one) is vendors who heard about ACPI, but haven't enough knowledge about the process. -- With Best Regards, Andy Shevchenko