On Wed, Jun 04, 2025 at 11:12:53AM +0200, Rafael J. Wysocki wrote: > On Wed, Jun 4, 2025 at 2:18 AM Ricardo Neri > <ricardo.neri-calderon@xxxxxxxxxxxxxxx> wrote: > > > > The bootstrap processor uses acpi_wakeup_cpu() to indicate to firmware that > > it wants to boot a secondary CPU using a mailbox as described in the > > Multiprocessor Wakeup Structure of the ACPI specification. > > > > The platform firmware may implement the mailbox as described in the ACPI > > specification but enumerate it using a DeviceTree graph. An example of > > this is OpenHCL paravisor. > > > > Move the code used to setup and use the mailbox for CPU wakeup out of the > > ACPI directory into a new smpwakeup.c file that both ACPI and DeviceTree > > can use. > > > > No functional changes are intended. > > > > Co-developed-by: Yunhong Jiang <yunhong.jiang@xxxxxxxxxxxxxxx> > > Signed-off-by: Yunhong Jiang <yunhong.jiang@xxxxxxxxxxxxxxx> > > Signed-off-by: Ricardo Neri <ricardo.neri-calderon@xxxxxxxxxxxxxxx> > > --- > > Changes since v3: > > - Create a new file smpwakeup.c instead of relocating it to smpboot.c. > > (Rafael) > > > > Changes since v2: > > - Only move to smpboot.c the portions of the code that configure and > > use the mailbox. This also resolved the compile warnings about unused > > functions that Michael Kelley reported. > > - Edited the commit message for clarity. > > > > Changes since v1: > > - None. > > --- > > arch/x86/Kconfig | 7 ++++ > > arch/x86/kernel/Makefile | 1 + > > arch/x86/kernel/acpi/madt_wakeup.c | 76 ---------------------------------- > > arch/x86/kernel/smpwakeup.c | 83 ++++++++++++++++++++++++++++++++++++++ > > 4 files changed, 91 insertions(+), 76 deletions(-) > > > > diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig > > index cb0f4af31789..82147edb355a 100644 > > --- a/arch/x86/Kconfig > > +++ b/arch/x86/Kconfig > > @@ -1113,6 +1113,13 @@ config X86_LOCAL_APIC > > depends on X86_64 || SMP || X86_UP_APIC || PCI_MSI > > select IRQ_DOMAIN_HIERARCHY > > > > +config X86_MAILBOX_WAKEUP > > + def_bool y > > + depends on OF || ACPI_MADT_WAKEUP > > At this point the dependency on OF is premature. IMV it should be > added in a later patch. I see your point. Sure, I will the dependency in a later patch.
