Re: still in 3.9-rc6 - fan speed at 100% after suspend/resume regression

From: Eduardo Valentin
Date: Mon Apr 15 2013 - 09:30:53 EST


On 13-04-2013 19:46, Ville Syrjälä wrote:
On Sun, Apr 14, 2013 at 12:44:00AM +0300, Ville Syrjälä wrote:
On Sat, Apr 13, 2013 at 10:02:21AM -0600, Jake Edge wrote:
Hi Zhang Rui,

The problem reported in https://lkml.org/lkml/2012/12/4/428 (and
incorrectly attributed to a suspend patch by me here:
https://lkml.org/lkml/2013/4/12/314) still exists in 3.9-rc6, at least
for my HP/Compaq 2510p laptop. It appeared that some folks were seeing
some improvement with earlier 3.9-rcs (?) but I just tested rc6 and saw
substantially the same behavior. This all works fine in <= 3.6.11.

A brief recap: after resuming, the fan on the laptop spins up to full
speed and stays there. "temp6" in "acpitz-virtual-0" (as shown by
"sensors") is 100°C and stays there, which is presumably what is making
the fan stay on.

My HP Compaq NC6000 exhibits slightly different behaviour. For me the
reported temp is always accurate but the trip points get out of sync
with the actual temperature.

With 3.7 I saw two different kinds of problems coming out of resume.
In one case the fan stays on until the temp rises high enough to get the
trip points back into sync. In the other case the fan goes off, and stays
off even when the temperature rises above the highest active trip point,
but it does appear to get back into sync when the temp starts to come back
down. I think the difference might stem from resuming when the laptop has
cooled down fully vs. when it's still warm.

I also just tried 3.9-rc6, and that one appears to behave differently
to 3.7, but still wrong. There the fan goes off after resume, and comes
back on as the temperature rises, but it never slows back down after
that.

I'll try to collect some more detailed dumps of all three cases.

I filed a new bug and attached all my logs:
https://bugzilla.kernel.org/show_bug.cgi?id=56591



Rui, does this one gets fixed with your series to make the parts of thermal framework to be glued together in a single module? Or does this one is a new bug?

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/