Re: [PATCH 1/3] ACPI autoloading - ACPICA adjustance

[Thomas Renninger <trenn@suse.de>]

On Tue, 2007-07-03 at 03:58 -0400, Len Brown wrote:
> On Sunday 17 June 2007 16:24, Thomas Renninger wrote:
> > Define standardized HIDs ...
> 
> > Index: linux-2.6.22-rc4/include/acpi/acpi_drivers.h
> > ===================================================================
> > --- linux-2.6.22-rc4.orig/include/acpi/acpi_drivers.h
> > +++ linux-2.6.22-rc4/include/acpi/acpi_drivers.h
> > @@ -34,16 +34,33 @@
> >  #define ACPI_BUS_COMPONENT		0x00010000
> >  #define ACPI_SYSTEM_COMPONENT		0x02000000
> >  
> > -/* _HID definitions */
> > +/*
> > +  _HID definitions
> > +  HIDs starting with 'X' are self defined here and are used for devices where
> > +  a HID does not exist in spec.
> > +  HIDs must conform to (ACPI spec - 6.1.4):
> >  
> > -#define ACPI_POWER_HID			"power_resource"
> > +  Numeric 32-bit compressed EISA type ID are not supported (never seen one as
> > +  an ACPI HID?)
> > +
> > +  A valid PNP ID must be of the form AAA#### where A is an uppercase letter
> > +  and # is a hex digit.
> > +  A valid ACPI ID must be of the form ACPI#### where # is a hex digit.
> > +
> > +  and our self-defined: XAAA#### where A is an uppercase letter and # is a hex
> > +  digit.
> 
> What was the criteria for inventing the XAAA#### format?
> It doesn't comply with the PNP standard format above, since
> it starts with four letters instead of three.
Yes, it still violates the standard, intend was to not do so.
And to stick to the 8/9 bytes.
The X has the same meaning as your suggestion below, starting with LNX,
it should be visible that it is an out of spec, linux only hid.
> 
> I don't think that "XVID0001" will be more meaningful than "video"
> when seen in sysfs...
> 
> > +*/
> > +
> > +#define ACPI_POWER_HID			"XPWR0001"
> >  #define ACPI_PROCESSOR_HID		"ACPI0007"
> > -#define ACPI_SYSTEM_HID			"acpi_system"
> > -#define ACPI_THERMAL_HID		"thermal"
> > -#define ACPI_BUTTON_HID_POWERF		"button_power"
> > -#define ACPI_BUTTON_HID_SLEEPF		"button_sleep"
> > -#define ACPI_VIDEO_HID			"video"
> > -#define ACPI_BAY_HID			"bay"
> > +#define ACPI_SYSTEM_HID			"XSYS0001"
> > +#define ACPI_THERMAL_HID		"XTHM0001"
> > +#define ACPI_BUTTON_HID_POWERF		"XPBT0001"
> > +#define ACPI_BUTTON_HID_SLEEPF		"XSBT0001"
> > +#define ACPI_VIDEO_HID			"XVID0001"
> > +#define ACPI_BAY_HID			"XBAY0001"
> > +
> 
> If the criteria is to stay within an 8 character id length,
> I'd rather spell a word with capital letters, and stick
> in 0's where we don't need more letters:
> 
> POWER000
> SYSTEM00
> THERMAL0
> PWRBTN00
> SLEEPBTN
> VIDEO000
> DRIVEBAY
> 
> Or when we discussed this before, we though about not conflicting
> with a standard (or vendor standard) name by becoming the Linux vendor --
> say using  the "LNX" prefix, and trying to spell something useful
> in the remaining 5 characters. eg.
> 
> LNXPOWER
> LNXSYSTM
> LNXTHERM
> LNXSLPBN
> LNXPWRBN
> LNXVIDEO
> LNXIOBAY

Don't know what is better, the first solution looks nicer...
Just tell me the final suggestion and I redo the patch.

    Thomas

PS: What is your general impression of the patches, will you add the
next series to acpi-test?
Shall I wait for some other patch(es) and then rebase, this one touches
nearly every file in /drivers/acpi/*.c and could end up in some merge
work for me or others.