On 21.01.2021 19:47, Julien Grall wrote: > On 21/01/2021 10:24, Jan Beulich wrote: >> On 21.01.2021 10:56, Julien Grall wrote: >>> On 21/01/2021 09:43, Jan Beulich wrote: >>>> On 21.01.2021 10:30, Michal Orzel wrote: >>>>> --- a/xen/include/asm-arm/early_printk.h >>>>> +++ b/xen/include/asm-arm/early_printk.h >>>>> @@ -10,6 +10,7 @@ >>>>> #ifndef __ARM_EARLY_PRINTK_H__ >>>>> #define __ARM_EARLY_PRINTK_H__ >>>>> >>>>> +#include <xen/page-size.h> >>>>> >>>>> #ifdef CONFIG_EARLY_PRINTK >>>> >>>> ... a respective Kconfig setting, i.e. it's not like I simply >>>> failed to enable it. >>> >>> EARLY_PRINTK is defined in arch/arm/Kconfig.debug and is selected when >>> you specify the UART to use. >>> >>> Assuming you are only build testing, you could add the following for >>> testing EARLY_PRINTK: >>> >>> CONFIG_DEBUG=y >>> CONFIG_EARLY_UART_CHOICE_8250=y >>> CONFIG_EARLY_UART_8250=y >>> CONFIG_EARLY_PRINTK=y >>> CONFIG_EARLY_UART_BASE_ADDRESS= >>> CONFIG_EARLY_UART_8250_REG_SHIFT=0 >>> CONFIG_EARLY_PRINTK_INC="debug-8250.inc" >> >> Ah yes, this works, thanks. The "optional" on the choice isn't >> very helpful I suppose, because when going from an existing >> .config one would neither find a setting presently turned off >> in that .config, nor will there be a prompt. > > Do you have a suggestion how the "choice" can appear in the .config?
Drop the "optional" attribute from it? Of course I ask without any knowledge on why it may have been put there in the first place. Jan
