On 31.10.2024 16:13, [email protected] wrote: > On Thu, 2024-10-31 at 11:29 +0100, Michal Orzel wrote: >> >> >> On 31/10/2024 10:34, Jan Beulich wrote: >>> >>> >>> On 30.10.2024 14:14, Oleksii Kurochko wrote: >>>> Introduce ic_preinit() in the common codebase, as it is not >>>> architecture-specific and can be reused by both PPC and RISC-V. >>>> This function identifies the node with the interrupt-controller >>>> property >>>> in the device tree and calls device_init() to handle >>>> architecture-specific >>>> initialization of the interrupt controller. >>>> >>>> Additionally, rename gic_acpi_preinit() to ic_acpi_preinit() as >>>> it is used >>>> by ic_preinit(), while keeping it defined in architecture- >>>> specific as this >>>> part is architecture-specific. In case if CONFIG_ACPI=n a stub >>>> for >>>> ic_acpi_preinit() is provided. To declaration/defintion of >>>> ic_acpi_preint() >>>> is added `inline` to deal with the compilation issue: >>>> error: 'ic_acpi_preinit' defined but not used [-Werror=unused- >>>> function] >>>> >>>> Make minor adjustments compared to the original ARM >>>> implementation of >>>> gic_dt_preinit(): >>>> - Remove the local rc variable in gic_dt_preinit() since it is >>>> only used once. >>>> - Change the prefix from gic to ic to clarify that the function >>>> is not >>>> specific to ARM’s GIC, making it suitable for other >>>> architectures as well. >>>> >>>> Signed-off-by: Oleksii Kurochko <[email protected]> >>>> --- >>>> Changes in v2: >>>> - Revert changes connected to moving of gic_acpi_preinit() to >>>> common code as >>>> it isn't really architecture indepent part. >>>> - Update the commit message. >>>> - Move stub of ic_acpi_preinit() to <asm-generic/device.h> for >>>> the case when >>>> CONFIG_ACPI=n. >>>> --- >>>> xen/arch/arm/gic.c | 45 +------------------------- >>>> ----- >>>> xen/arch/arm/setup.c | 3 ++- >>>> xen/common/device.c | 46 >>>> ++++++++++++++++++++++++++++++++ >>>> xen/include/asm-generic/device.h | 10 +++++++ >>>> 4 files changed, 59 insertions(+), 45 deletions(-) >>> >>> Looks largely okay to me now, with a question and a nit at the >>> bottom. The >>> question is mainly to Arm folks, where the code is coming from, and >>> DT >>> maintainers: Is a file named device.c really an appropriate "home" >>> for >>> IC-related code? If IC is a common thing in the DT world, would >>> such code >>> maybe better live under common/device-tree/? >> I think we will have more interrupt related dt stuff common, so it >> would make sense to create >> device-tree/irq.c or intc.c. > What about the part of the code in common/device.c: > void __init ic_preinit(void) > { > if ( acpi_disabled ) > ic_dt_preinit(); > else > ic_acpi_preinit(); > } > > Should it be also moved to device-tree/intc.c even ic_acpi_preinit() is > used?
Maybe best left to each arch? device.[ch] aren't a good home for this, just like they aren't for the actual DT implementation. Jan
