Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 01 Nov 2005 11:06:14 -0800
From:      Nate Lawson <nate@root.org>
To:        Mathieu Prevot <bsdhack@club-internet.fr>
Cc:        freebsd-acpi@FreeBSD.org
Subject:   Re: ACPI and 3.0 specification
Message-ID:  <4367BCA6.5050609@root.org>
In-Reply-To: <78F7D8FC-B5AA-4723-8336-E60F873D9414@club-internet.fr>
References:  <78F7D8FC-B5AA-4723-8336-E60F873D9414@club-internet.fr>

next in thread | previous in thread | raw e-mail | index | archive | help
Mathieu Prevot wrote:
> On Sept. 2004 the 3.0 specification has been released. Change  
> description is:
> 
> Major specification revision. General configuration enhancements.  
> Inter-Processor power, performance, and throttling state dependency  
> support added. 

We already support some of the processor enhancements through 
cpufreq(4).  acpi_cpu was ahead of the 3.0 spec in implementing SMP Cx 
support.

A few laptops are starting to use the _{C,P,T}SD methods for 
dependencies but they still work with acpi 2.0 compatibility also.

> Support for > 256 processors added. NUMA Distancing  
> support added. PCI Express support added. SATA support added.

We don't support locality and other things from ACPI 2.0.  PCI express 
and SATA are supported by device-specific drivers.

> Ambient  
> Light Sensor and User Presence device support added. Thermal model  
> extended beyond processor-centric support.

Never seen this on any system.

> Does implementing these specifications (and maybe also 2.0) correctly  
> would help making fbsd more fluid with talking to hardware including  
> most recent MB ?
> How often is motherboards' AML critic ?
> How many of you are working on it (how long / month) and how  (partly / 
> exclusively) ?
 >
> I am thinking about working on it, depending on the time I can spend  
> (still student in physics (graduate), preparing also PhD in  
> nanomaterials and piano concert for 2006).

acpi 3.0 adds very little useful stuff unless you're interested in large 
NUMA machines.  We'd be better off implementing more support for those 
systems in the main kernel and then acpi, not the other way around.

-- 
Nate



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4367BCA6.5050609>