[ltp] possible fan firmware bug after WLAN was active on X61s

Henrique de Moraes Holschuh linux-thinkpad@linux-thinkpad.org
Thu, 17 Jul 2008 09:05:21 -0300


On Wed, 16 Jul 2008, Thomas Kahle wrote:
> I heard some rumors that the list of changes for new Bios versions is
> not complete, so maybe they also changed something with the fan in newer
> versions.

That's not a rumor.  It is first-hand verified and confirmed information.

> What convinced me was the availability of the "watchdog", as soon as the
> script does not set the fan-levels for 5 seconds, say, the BIOS retakes
> control.

Yeah, I added it because otherwise doing fan-control in userspace would be
too damn dangerous.  At least kernel threads never get OMM-killed.

Which doesn't mean it will be safe if your machine hangs.  Never leave a
thinkpad in non-EC fan control unattended (unless it is sleeping or
somesuch).   AFAIK, the hardware reboot watchdog ("Intel TCO watchdog") of
the chipset don't work with the ThinkPad BIOS, the thing just hangs the box
HARD if it activates, so you can't even have that as a second anti-hang
measure.

> Furthermore I extensively monitored the temperatures and the
> miniPCI(wlan)-temperature is THE SAME with the script on or BIOS
> control. (Since the fan does not cool the miniPCI). For all other
> sensors I left the threshholds at conservative values that are easily
> reached with BIOS control too.
>
> For this two reasons I decided it is safe enough for me.

If you are reproducing the EC behaviour, it indeed should be safe.

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh