Re: [Devel] GPE blocks in reduced hardware mode

[Al Stone <ahs3 at redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 5993 bytes --]

On 12/04/2013 08:51 AM, Moore, Robert wrote:
> Some text pulled from the ACPI 5.0 Spec:
>
>
> 5.6 ACPI Event Programming Model
> The ACPI event programming model is based on the SCI interrupt and General-Purpose Event (GPE) register. ACPI provides an extensible method to raise and handle the SCI interrupt, as described in this section.
> Hardware-reduced ACPI platforms (Section 4.1) use GPIO Interrupt Connections to signal ACPI Events, described in Section 5.6.5.
>
>
> 5.6.1 ACPI Event Programming Model Components
> The components of the ACPI event programming model are the following:
> * OSPM
> * FADT
> * PM1a_STS, PM1b_STS and PM1a_EN, PM1b_EN fixed register blocks
> * GPE0_BLK and GPE1_BLK register blocks
> * GPE register blocks defined in GPE block devices
>
>
> 5.6.5 GPIO-signaled ACPI Events
> On Hardware-reduced ACPI platforms, ACPI events are signaled when a GPIO Interrupt is received by OSPM,
>
>
> 7.2.13 PRW (Power Resources for Wake)
> For HW-Reduced ACPI platforms, the GpeInfo structure is ignored by OSPM.
>
>
>>> Hardware-reduced ACPI platforms (Section 4.1) use GPIO Interrupt Connections to signal ACPI Events.
>
> I think this sentence says that on hardware-reduced ACPI platforms, there are no GPEs, period.
> Bob

Thanks for the pointers; I'm afraid I can't add anything to that
list -- it matches what I came up with when I scanned again last
night.

I agree with you that the spec is saying are no GPEs in hardware
reduced, but I think it is implied, versus it being stated explicitly.
And that's where I get a bit puzzled.

I'll take this up with the ASWG and see if I can get some sort of
unambiguous wording added in.

>> -----Original Message-----
>> From: Devel [mailto:devel-bounces(a)acpica.org] On Behalf Of Moore, Robert
>> Sent: Tuesday, December 03, 2013 10:28 PM
>> To: Al Stone; devel(a)acpica.org
>> Subject: Re: [Devel] GPE blocks in reduced hardware mode
>>
>>
>>
>>> -----Original Message-----
>>> From: Al Stone [mailto:ahs3(a)redhat.com]
>>> Sent: Tuesday, December 03, 2013 5:39 PM
>>> To: Moore, Robert; devel(a)acpica.org
>>> Subject: Re: [Devel] GPE blocks in reduced hardware mode
>>>
>>> On 12/03/2013 05:18 PM, Moore, Robert wrote:
>>>> The first thing is that there are no GPE block devices in the field,
>>> never have been. I don't think ACPICA actually supports them fully. It
>>> is a bit of an "orphan" feature of ACPI.
>>>
>>> I see.  Yeah, I did find GPE block device support confusing in the code.
>>> It seems like the infrastructure to handle them is present, but there
>>> is no code to handle the "ACPI0006" ID anywhere I can find (that may
>>> be a Linux limitation, though).
>>
>>
>> Correct. There needs to be a linux driver that is loaded upon detection of
>> an _HID of ACPI0006. This does not exist. Thus, there is no support for
>> this feature in Linux.
>>
>> The GPE block device was a bit of a fantasy that never materialized in any
>> hardware that I know of. And one could certainly conclude with some high
>> degree of probability that nobody is going to implement custom hardware to
>> support a GPE block device on a (ahem) "hardware reduced platform".
>>
>> Bob
>>
>>
>>
>>
>>>
>>>> I'll have to take a look closer at the spec for an exact reference,
>>>> but
>>> basically one of the things that ACPI_REDUCED_HARDWARE means is no
>>> GPEs, IIRC.
>>>
>>> Right -- that's what ACPI_REDUCED_HARDWARE means in the ACPICA code;
>>> that seems pretty clear.  I just can't seem to find anything that
>>> clear in the spec :/...
>>>
>>> I'll continue to look in the spec but if you find something before I
>>> do, do let me know.
>>>
>>> Thanks!
>>>
>>>>> -----Original Message-----
>>>>> From: Devel [mailto:devel-bounces(a)acpica.org] On Behalf Of Al Stone
>>>>> Sent: Tuesday, December 03, 2013 2:38 PM
>>>>> To: devel(a)acpica.org
>>>>> Subject: [Devel] GPE blocks in reduced hardware mode
>>>>>
>>>>> Howdy.
>>>>>
>>>>> I've been trying to do some clean-up on the Linux ACPI driver with
>>>>> regard to the reduced hardware profile introduced in ACPI 5.0.
>>>>>
>>>>> In thinking about GPE block devices, I was looking through the
>>>>> functions in source/components/events/evxfgpe.c and noticed that
>>>>> they are all #ifdef'd out for ACPI_REDUCED_HARDWARE.  When I look
>>>>> back through the specification, however, I cannot find any place
>>>>> where they
>>> are prohibited.
>>>>>
>>>>> Clearly, the gpe0/1 blocks in the FADT are not to be used; this is
>>>>> explicitly stated in section 5.2.9 of the spec.  Could someone
>>>>> please point me to a similar statement for a GPE block device?  I
>>>>> just cannot seem to find it.
>>>>>
>>>>> The closest thing I can find is section 9.10 implying that a GPE
>>>>> block device is an extension of the gpe0/1 blocks from the FADT,
>>>>> but there is no explicit statement there so it is ambiguous to me.
>>>>>
>>>>> Thanks in advance for any pointers; I'm just puzzled at this point
>>>>> and trying to understand how the conclusion came to be.
>>>>>
>>>>> --
>>>>> ciao,
>>>>> al
>>>>> -----------------------------------
>>>>> Al Stone
>>>>> Software Engineer
>>>>> Red Hat, Inc.
>>>>> ahs3(a)redhat.com
>>>>> -----------------------------------
>>>>> _______________________________________________
>>>>> Devel mailing list
>>>>> Devel(a)acpica.org
>>>>> https://lists.acpica.org/mailman/listinfo/devel
>>>
>>>
>>> --
>>> ciao,
>>> al
>>> -----------------------------------
>>> Al Stone
>>> Software Engineer
>>> Red Hat, Inc.
>>> ahs3(a)redhat.com
>>> -----------------------------------
>> _______________________________________________
>> Devel mailing list
>> Devel(a)acpica.org
>> https://lists.acpica.org/mailman/listinfo/devel


-- 
ciao,
al
-----------------------------------
Al Stone
Software Engineer
Red Hat, Inc.
ahs3(a)redhat.com
-----------------------------------