On Fri, 2 Oct 2026, at 12:31, Màxim Pedraza Padilla wrote: > Hi Jani, > > On Fri, 02 Oct 2026, Jani Nikula wrote: >> Data is not an interface. >> >> If these will be used more, perhaps it would be better to wrap access to >> them in functions? I think most of the time it will help with >> maintenance. >> >> And you can add stubs for CONFIG_ACPI_BGRT=n where they belong, >> i.e. include/linux/efi-bgrt.h instead of drm_splash.c like in patch 11 >> of this series. > > Agreed, and the same goes for your comment on patch 3. For v5 I would > replace the export with accessors in efi-bgrt.c, stubbed in > efi-bgrt.h for CONFIG_ACPI_BGRT=n: > > bool efi_bgrt_has_image(void); > phys_addr_t efi_bgrt_image_address(void); > size_t efi_bgrt_image_size(void); > u8 efi_bgrt_status(void); > u32 efi_bgrt_image_offset_x(void); > u32 efi_bgrt_image_offset_y(void); > > The splash client then drops its own wrappers and stubs, and the > "static inline" in drm_splash.c go as well. > > drivers/acpi/bgrt.c is the only other user of bgrt_tab and > bgrt_image_size. I can move it to the accessors too, so that both > variables become static, if that is wanted; otherwise they stay > global for it and are no longer exported. > > Ard, this changes the patch you acked, so I will not carry your > Acked-by unless the above works for you. Both variables become > __ro_after_init either way, as you asked. >
This sounds good to me - please cc me again on the result and I'll take another look.
