Re: [PATCH] x86: Bypass legacy PIC and PIT on ACPI hardware reduced platform

[Ingo Molnar <mingo@kernel.org>]

* Li, Aubrey <aubrey.li@linux.intel.com> wrote:

> On 2015/3/4 13:31, Ingo Molnar wrote:
> > 
> > * Li, Aubrey <aubrey.li@linux.intel.com> wrote:
> > 
> >> On 2015/3/4 13:08, Ingo Molnar wrote:
> >>>
> >>> * Li, Aubrey <aubrey.li@linux.intel.com> wrote:
> >>>
> >>>> On ACPI hardware reduced platform, the legacy PIC and PIT may not be
> >>>> initialized even though they may be present in silicon. Touching
> >>>> these legacy components causes unexpected result on system.
> >>>>
> >>>> On Bay Trail-T(ASUS-T100) platform, touching these legacy components
> >>>> blocks platform hardware low idle power state(S0ix) during system suspend.
> >>>> So we should bypass them on ACPI hardware reduced platform.
> >>>>
> >>>> Suggested-by: Arjan van de Ven <arjan@linux.intel.com>
> >>>> Signed-off-by: Li Aubrey <aubrey.li@linux.intel.com>
> >>>> Cc: Len Brown <len.brown@intel.com>
> >>>> Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> >>>> ---
> >>>>  arch/x86/kernel/irqinit.c | 6 +++++-
> >>>>  arch/x86/kernel/time.c    | 3 ++-
> >>>>  2 files changed, 7 insertions(+), 2 deletions(-)
> >>>>
> >>>> diff --git a/arch/x86/kernel/irqinit.c b/arch/x86/kernel/irqinit.c
> >>>> index 70e181e..9a64cc3 100644
> >>>> --- a/arch/x86/kernel/irqinit.c
> >>>> +++ b/arch/x86/kernel/irqinit.c
> >>>> @@ -75,7 +75,11 @@ void __init init_ISA_irqs(void)
> >>>>  #if defined(CONFIG_X86_64) || defined(CONFIG_X86_LOCAL_APIC)
> >>>>  	init_bsp_APIC();
> >>>>  #endif
> >>>> -	legacy_pic->init(0);
> >>>> +	if (acpi_gbl_reduced_hardware) {
> >>>> +		pr_info("Using NULL legacy PIC\n");
> >>>> +		legacy_pic = &null_legacy_pic;
> >>>> +	} else
> >>>> +		legacy_pic->init(0);
> >>>>  
> >>>>  	for (i = 0; i < nr_legacy_irqs(); i++)
> >>>>  		irq_set_chip_and_handler(i, chip, handle_level_irq);
> >>>> diff --git a/arch/x86/kernel/time.c b/arch/x86/kernel/time.c
> >>>> index 25adc0e..5ba94fa 100644
> >>>> --- a/arch/x86/kernel/time.c
> >>>> +++ b/arch/x86/kernel/time.c
> >>>> @@ -14,6 +14,7 @@
> >>>>  #include <linux/i8253.h>
> >>>>  #include <linux/time.h>
> >>>>  #include <linux/export.h>
> >>>> +#include <linux/acpi.h>
> >>>>  
> >>>>  #include <asm/vsyscall.h>
> >>>>  #include <asm/x86_init.h>
> >>>> @@ -76,7 +77,7 @@ void __init setup_default_timer_irq(void)
> >>>>  /* Default timer init function */
> >>>>  void __init hpet_time_init(void)
> >>>>  {
> >>>> -	if (!hpet_enable())
> >>>> +	if (!hpet_enable() && !acpi_gbl_reduced_hardware)
> >>>>  		setup_pit_timer();
> >>>>  	setup_default_timer_irq();
> >>>>  }
> >>>
> >>> So the whole acpi_gbl_reduced_hardware flaggery sucks as it mixes 
> >>> various hardware drivers that have little relation to each other...
> >>>
> >>> Instead of having a proper platform init this flag hooks into various 
> >>> drivers and generic code, such as the efi reboot and shutdown code, 
> >>> and now the generic irq init code.
> >>>
> >>> For this IRQ init problem, why not add a proper callback to 
> >>> x86_platform_ops, define your own IRQ init function, initialize it in 
> >>> your platform init sequence and let it be called? That solves it 
> >>> without creating an ugly mix of different platform methods.
> >>>
> >>> For the EFI shutdown case, what's wrong with setting your own 
> >>> pm_power_off handler like most of the other platforms are doing? Plus 
> >>> the EFI code in drivers/firmware/efi/reboot.c should probably only set 
> >>> the shutdown handler if pm_power_off is still NULL.
> >>
> >> I think our goal is to make the code as generic as possible for all 
> >> x86 platform, rather than creating a new x86 branch, I added Alan 
> >> Cox for this strategy discussion.
> >>
> >> Do you have any inputs for the patch itself?
> > 
> > Other than that the patch is unacceptable for an upstream merge in its 
> > current form for the reason I mentioned? No.
> 
> So you are suggesting we extend a new x86 platform branch and 
> override the x86_platform and pm_power_off and reboot, like what 
> intel_mid does?

Well, what I suggested above was to add an IRQ init method to 
x86_platform (and make use of it on your platform), and to
use the existing pm_power_off method for the reboot quirk.

Using 'acpi_gbl_reduced_hardware' flag outside the ACPI code
is a mistake.

Thanks,

	Ingo