[Debian-eeepc-devel] Bug#559578: Bug#559578: Fwd: Bug#559578: eeepc-acpi-scripts: EeePC 701 freezes with garbled screen while booting

[Alan Jenkins]

On 12/10/09, Darren Salt <linux at youmustbejoking.demon.co.uk> wrote:
> I demand that Alan Jenkins may or may not have written...
>
>> On 12/9/09, Darren Salt <linux at youmustbejoking.demon.co.uk> wrote:
>>> I demand that Alan Jenkins may or may not have written...
>>> [snip]
>>>> I think the kernel should disable SHE by default on the 701 models.  I
>>>> don't mind if it provides a "force_cpufv" option with an appropriate
>>>> warning.
>>> And I have no problem with modifying eeepc-acpi-scripts accordingly:
>>> adding documentation ("How to break your 701"), and generating a warning
>>> and not altering the SHE setting if this option is missing and the script
>>> is being run on a 701.
>>> We'll need to support current and older kernels for a while yet...
>>> [snip]
>>>> I don't think it should be left to userspace to get this right.
>>> Agreed.
>>> [snip]
>
>> Ok. If we use a normal module option, the kernel won't be able to provide
>> a warning when it is set at run-time.
>
> I meant the init script for eeepc-acpi-scripts, but having the kernel output
> a warning wouldn't be a bad thing.
>
>> Setting module parameters under /sys/module/*/parameters just sets a
>> variable; it doesn't call into the module.
>
> True...
>
>> So I guess we want "force_enable_cpufv" as a new sysfs attribute.
>
> Probably, yes. I was thinking along the lines of "you want this, you make
> sure that the appropriate line is present in a file in /etc/modprobe.d", but
> I think that this will be more supportable in that it's easier for your
> average user to test without committing to the change.

My reasoning was

a) You're going to add a DMI check in eeepc-acpi-scripts (for existing
kernels which don't check the model number).
b) There are at least two developers who used SHE on the 701 despite
it not being officially supported.  I therefore anticipate that
eeepc-acpi-scripts will include a SHE_FORCE_ENABLE option to skip the
userspace DMI check.
c) For consistency, SHE_FORCE_ENABLE should also cause the DMI check
to be skipped on newer kernels (by writing to the sysfs attribute).

Thanks
Alan