[ltp] [patch 11/12][2.6.18] ibm-acpi: workaround for EC 0x2f initialization bug

[Shem Multinymous]

Hi,

On 10/23/06, Henrique de Moraes Holschuh <hmh@hmh.eng.br> wrote:
> There is another option.  I can overengineer the thing.  I can read all
> thermal sensors, and if all sensors are within some safe range (say, below
> 60 C), kick the fan to mode auto.  If they are outside that range, kick the
> fan to mode 0x07.

That's dangerous; for one, nothing guarantees we see all sensors. It's
best done by userspace, if at all.

> What is your vote on the issue, as a T43 owner?

Don't lose information. Output "uninitialized" and let userspace deal
with it as it wishes.


> > >                if (unlikely(!acpi_ec_read(fan_rpm_offset, &lo) ||
> > >                          !acpi_ec_read(fan_rpm_offset + 1, &hi)))
> > >                        return -EIO;

Sorry, that's perfectly fine. I read it wrong.


> I am hooking the watchdog to the procfs (and later, sysfs) handlers and not
> from the generic functions, so they won't clash.
>
> Do you think the watchdog should be hooked from the generic functions
> instead?  I could do that, as well... might be a better idea, actually, but
> if I do that we lose the ability to call them without affecting the
> watchdog.

No, you're right, it's better this way. The watchdog should only be
reset by explicit writes from userspace, so procfs and sysfs are the
right place for the hook.

  Shem