Re: [PATCH 1/2] ACPI / PM: Use Low Power S0 Idle on more systems

[Rafael J. Wysocki]

On Wed, Jan 10, 2018 at 6:38 PM,  <Mario.Limonciello@xxxxxxxx> wrote:
>
>
>> -----Original Message-----
>> From: platform-driver-x86-owner@xxxxxxxxxxxxxxx [mailto:platform-driver-x86-
>> owner@xxxxxxxxxxxxxxx] On Behalf Of Rafael J. Wysocki
>> Sent: Wednesday, January 10, 2018 6:26 AM
>> To: Linux ACPI <linux-acpi@xxxxxxxxxxxxxxx>
>> Cc: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>; Darren Hart
>> <dvhart@xxxxxxxxxxxxx>; LKML <linux-kernel@xxxxxxxxxxxxxxx>; Linux PM <linux-
>> pm@xxxxxxxxxxxxxxx>; Platform Driver <platform-driver-x86@xxxxxxxxxxxxxxx>;
>> Valentin Manea <valy@xxxxxx>
>> Subject: [PATCH 1/2] ACPI / PM: Use Low Power S0 Idle on more systems
>>
>> From: Rafael J. Wysocki <rafael.j.wysocki@xxxxxxxxx>
>>
>> Some systems don't support the ACPI_LPS0_ENTRY and ACPI_LPS0_EXIT
>> functions in their Low Power S0 Idle _DSM, but still expect EC
>> events to be processed in the suspend-to-idle state for power button
>> wakeup (among other things) to work.  Surface Pro3 turns out to be
>> one of them.
>>
>> Fortunately, it still provides Low Power S0 Idle _DSM with the screen
>> on/off functions supported, so modify the ACPI suspend-to-idle to use
>> the Low Power S0 Idle code path for all systems supporting the
>> ACPI_LPS0_ENTRY and ACPI_LPS0_EXIT or the ACPI_LPS0_SCREEN_OFF and
>> ACPI_LPS0_SCREEN_ON functions in their Low Power S0 Idle _DSM.
>>
>> Potentially, that will cause more systems to use suspend-to-idle by
>> default, so some future corrections may be necessary if it leads
>> to issues, but let it remain more straightforward for now.
>>
>> Link: https://bugzilla.kernel.org/show_bug.cgi?id=198389
>> Reported-by: Valentin Manea <valy@xxxxxx>
>> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@xxxxxxxxx>
>> ---
>>  drivers/acpi/sleep.c |    6 ++++--
>>  1 file changed, 4 insertions(+), 2 deletions(-)
>>
>> Index: linux-pm/drivers/acpi/sleep.c
>> ===================================================================
>> --- linux-pm.orig/drivers/acpi/sleep.c
>> +++ linux-pm/drivers/acpi/sleep.c
>> @@ -707,7 +707,8 @@ static const struct acpi_device_id lps0_
>>  #define ACPI_LPS0_ENTRY              5
>>  #define ACPI_LPS0_EXIT               6
>>
>> -#define ACPI_S2IDLE_FUNC_MASK        ((1 << ACPI_LPS0_ENTRY) | (1 <<
>> ACPI_LPS0_EXIT))
>> +#define ACPI_LPS0_SCREEN_MASK        ((1 << ACPI_LPS0_SCREEN_OFF) | (1 <<
>> ACPI_LPS0_SCREEN_ON))
>> +#define ACPI_LPS0_S2I_MASK   ((1 << ACPI_LPS0_ENTRY) | (1 << ACPI_LPS0_EXIT))
>>
>>  static acpi_handle lps0_device_handle;
>>  static guid_t lps0_dsm_guid;
>> @@ -910,7 +911,8 @@ static int lps0_device_attach(struct acp
>>       if (out_obj && out_obj->type == ACPI_TYPE_BUFFER) {
>>               char bitmask = *(char *)out_obj->buffer.pointer;
>>
>> -             if ((bitmask & ACPI_S2IDLE_FUNC_MASK) ==
>> ACPI_S2IDLE_FUNC_MASK) {
>> +             if ((bitmask & ACPI_LPS0_S2I_MASK) == ACPI_LPS0_S2I_MASK ||
>> +                 (bitmask & ACPI_LPS0_SCREEN_MASK) ==
>> ACPI_LPS0_SCREEN_MASK) {
>>                       lps0_dsm_func_mask = bitmask;
>>                       lps0_device_handle = adev->handle;
>>                       /*
>
> In making this change I believe you'll need to cache the values that you found from the
> function mask to test them later too.

But that's what lps0_dsm_func_mask is for if I understand you correctly.

> Here:
> https://github.com/torvalds/linux/blob/master/drivers/acpi/sleep.c#L943
>
> This is because later on both ACPI_LPS0_SCREEN_OFF and ACPI_LPS0_ENTRY are called
> whether or not they both exist.

No, that's not the case.

acpi_sleep_run_lps0_dsm() checks if the given function is there in the
mask returned by function 0 and it doesn't evaluate the _DSM
otherwise.

Thanks,
Rafael