Problems with acpi and dmidecode

[Richard Hughes]

On Tue, 2005-10-11 at 00:19 +0200, Danny Kukawka wrote:
> On Monday 10 October 2005 23:38, Richard Hughes wrote:
> > My experimental patches failed to work to fix the broken BIOS of a
> > COMPAQ EVO N160, for reasons that I now realise:
> >
> > smbios.* is set after acpi_synthesize_hotplug_events is called.
> >
> > My dmidecode idea made the BIOS correction check in acpi.c.
> >
> > There are two ways to fix this:
> >
> > 1. Call hal_util_helper_invoke ("hald-probe-smbios") before we call
> > acpi_synthesize_hotplug_events (which will work, and hal-probe-smbios
> > should just exit cleanly with powermacs, as dmidecode isn't found..)
> >
> > 2. We can do the ACPI BIOS corrections in osspec.c (not ideal, as it
> > leaks the acpi specific bits into a general module)
> >
> 
> 3. Maybe use fdi file which merge the keys if match realted smbios.* keys and 
> add a key with stringlist of needed ACPI_QUIRK_NOT_* . This would also be 
> easyer to maintain, changeable by the user and more generic.

Never thought of using an fdi file, that's a good idea.

I'll have a think tomorrow.

Richard.