Re: [Gnewsense-dev] Re: 2.6.32-libre-lemote vs NetworkManager (race condition?)

[Wu Zhangjin]

On Wed, 2009-12-16 at 21:32 +0800, Wu Zhangjin wrote:
> On Wed, 2009-12-16 at 10:15 -0200, Alexandre Oliva wrote:
> > On Dec 15, 2009, Wu Zhangjin <address@hidden> wrote:
> > 
> > > Could you please pull the latest 2.6.32 branch? Just fixed some problems
> > > of the fan controlling interface, this one should work well with the
> > > fancontrol.
> > 
> > I did.  The fan controlling interface changes are in
> > 2.6.32.1-libre-lemote and 2.6.31.8-libre2-lemote (I noticed you
> > backported them to the 2.6.31-stable branch too).
> > 
> > Unfortunately, they don't seem to help.
> > 
> > sensors reports the temperature steadily rising and the fan stuck at
> > some 440 RPM, unlike the other laptop in which the fan gets up to some
> > 4800 RPM.
> > 
> > Furthermore, pwmconfig (that's supposed to create /etc/fancontrol) says
> > there aren't any pwm interfaces.
> > 
> > Regardless, taking over manual control of hwmon1/pwm1 and setting it to
> > 3, as you suggested, didn't change the fan speed at all.  It seems that
> > 439 RPM is its max speed ATM :-(
> > 
> 
> So, perhaps your fan have problems ;-(
> 
> > Unless you have other suggestions, it seems like the only way for this
> > machine to work reliably (at least until we can fix its fan) is to set
> > the CPU speed to the minimum.

If your fan is not broken, perhaps you can try to update your EC ROM?

The latest EC version is PQ1D26.

check the current version in kernel space:

$ cat /proc/cmdline

or get it in PMON command line:

$ set

If it is too old, please tell me, I will send the latest one to you.

> > 
> > 
> > BTW, is there any hope for auto-detecting machtype in the kernel, rather
> > than having it passed in from the boot command line?  I'm wondering if
> > the need for specifying it is a temporary stop-gap or something
> > long-term that userland must be adjusted for.  Auto-detection would be
> > highly desirable, but I don't know enough about this hardware to even
> > begin to tell whether there are any hard impediments to implementing it.
> > 
> 
> In the future, the machtype will be passed by BIOS(PMON) automatcially,
> there is no way to check the machtype currently. of course, perhaps we
> can preserve several KBs to save the information of the whole machine,
> and then kernel can access it later.
> 
> Regards,
>       Wu Zhangjin