2003-05-02 18:20:02

by Jeffrey W. Baker

[permalink] [raw]
Subject: Re: centrino

Here was my collected experience on a Centrino machine,
before I returned it in favor of an iBook:

The AGP worked in 2.4.21-pre but not in 2.5.68-mm2,
regardless of agp_try_unsupported=1. The Radeon 9000
Mobility in the Acer Travelmate is supported in XFree86
4.3.0. Don't know about the IBM.

The IDE DMA seemed to work fine in both 2.4.21 and 2.5.68.

Power management didn't work at all because ACPI is a sick
joke. In 2.4 ACPI does nothing, and in 2.5.68 it can put
the machine to sleep, but not wake it up. APM could almost
wake the system from sleep, but then it crashed immediately.
ACPI incorrectly reports the state of the system fans, the
battery, the battery charger, and the temperature sensor.
In other words, no part of it functions correctly . This is
either a problem with Centrino chipset in general or Acer
BIOS programming in particular.

CPU frequency scaling didn't work in either kernel but there
seem to be rumbles of reverse engineering going on with
that.

The wireless doesn't work in any kernel, and Intel have
stated specifically on their web site that they have no
plans for any future Linux driver for that device.

The Centrino package is altogether hostile to Linux.

-jwb


2003-05-03 16:08:07

by Alan

[permalink] [raw]
Subject: Re: centrino

> Power management didn't work at all because ACPI is a sick
> joke. In 2.4 ACPI does nothing, and in 2.5.68 it can put
> the machine to sleep, but not wake it up. APM could almost
> wake the system from sleep, but then it crashed immediately.
> ACPI incorrectly reports the state of the system fans, the
> battery, the battery charger, and the temperature sensor.
> In other words, no part of it functions correctly . This is
> either a problem with Centrino chipset in general or Acer
> BIOS programming in particular.

For some people centrino seems to work a little better so it may
be BIOS bugs too. The usual "buy something 6 months old" rule applies
to BIOS code as well as anything else

> CPU frequency scaling didn't work in either kernel but there
> seem to be rumbles of reverse engineering going on with
> that.

Intel are still being a pain about this. Quite why they think its
a secret is beyond me. Fortunately every other CPU vendor is being a
lot more positive.

> The wireless doesn't work in any kernel, and Intel have
> stated specifically on their web site that they have no
> plans for any future Linux driver for that device.

There are awkward issues there certainly.

> The Centrino package is altogether hostile to Linux.

Time may change that. The ATI/ALi chipset AMD laptops like the Compaq
z9xx were really Linux hostile. Now with the right stuff in place and
some things fixed (some working around their bugs, some fixing ours)
they work really well.

Centrino ultimately is a help - the laptop chipset space is
consolidating and there are less and less variations appearing. That
helps us because there is less to support. Its the same as with X and
IDE and other things nowdays - we get a lot less deeply weird hardware
to fight.

2003-05-03 20:16:42

by Anders Karlsson

[permalink] [raw]
Subject: Re: centrino

On Sat, 2003-05-03 at 16:21, Alan Cox wrote:

> > The wireless doesn't work in any kernel, and Intel have
> > stated specifically on their web site that they have no
> > plans for any future Linux driver for that device.
>
> There are awkward issues there certainly.

Alan, have you had any better luck talking to Intel than I had? At least
you are a well known person in the Linux environment and have a lot more
clout than when a complete nobody like me tries to get their attention.
Getting the wireless working would be a great step forward.

> > The Centrino package is altogether hostile to Linux.
>
> Time may change that. The ATI/ALi chipset AMD laptops like the Compaq
> z9xx were really Linux hostile. Now with the right stuff in place and
> some things fixed (some working around their bugs, some fixing ours)
> they work really well.
>
> Centrino ultimately is a help - the laptop chipset space is
> consolidating and there are less and less variations appearing. That
> helps us because there is less to support. Its the same as with X and
> IDE and other things nowdays - we get a lot less deeply weird hardware
> to fight.

Eventually things will work in favour of linux. It usually does. The
time it might take to get there is a variable factor though. Personally
I thought Centrino was a good idea. I normally don't like 'Chipzilla'
but in this instance I think they did a good thing.

Regards,

/Anders


Attachments:
signature.asc (198.00 B)
This is a digitally signed message part