[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: PowerMac G5 fans run out of control with kernel4.17.0-3-powerpc64 but not with 4.16.0-1-powerpc64



Hi Mark!

On Aug 25, 2018, at 8:52 PM, Mark G.B. <mark.balantzyan@gmail.com> wrote:

> Hello Rick and other PMac users,
>  
> What kind of Powermac are you running the 4.17 kernel on? 11,2? 11,2 btw, I believe is the G5 Quad-core. Even with your reply, I may not come to a definitive conclusion. I will have to do some research, and I certainly am not nearly as well-versed as those on the Debian team.
>  
> For reference, my own Powermac, an 7,2, does the EXACT same things, /etc/modules “windfarm_core” eliminates the tornado. How did you even find that fix? The origin of that fix may help us determine the backend-kernel functionality pointing porters in the right direction.
>  
> Regards,
> Mark

To answer your questions:

It’s a G5 version “7,3”, with a 2.0 GHz dual-core/single-chip CPU, with 4GB RAM and a 120GB hard disk.

As I mentioned in my original email to this list,
> Back in the old days of Linux kernel 3.x, the fix for wind tunnel fans was to load the “i2c_powermac" kernel module.  Nowadays, that module is built-in so it isn’t necessary (or even possible) to manually load it.  I checked the config files, and it is still built-in on both the 4.16 and the 4.17 kernels.

The pre 4.x kernel “i2c_powermac” fix came from an old email on this list — I don’t remember who it was from.  It seemed to be common knowledge on the list at the time — common enough that making the module built-in with the 4.x powerpc64 kernels was uncontroversial.

Then Dennis Clarke replied to my original email, mentioning that it might have something to do with “windfarm”.  I had noticed that keyword in the list of available modules, so looking at the output of “lsmod” was an obvious first-step.  When I noticed that “windfarm_core” (and a bunch of other “windfarm” modules that it depends on) were missing from the running 4.17 kernel, it seemed that doing “modload windfarm_core” was at least worth a try.  I did try it, and it worked — the fans quieted down immediately.  I put that knowledge together with the old information about “i2c_powermac” and gave it a try.

So the question we need to answer is:  What happened between 4.16.0-1-powerpc64 and 4.17.0-3-powerpc64 kernels that automatically loaded the “windfarm_core” module in 4.16 but not in 4.17?  And is the change a bug or a feature?

Enjoy!
Rick

> rbthomas@kmac:~$ cat /proc/cpuinfo
> processor	: 0
> cpu		: PPC970FX, altivec supported
> clock		: 2000.000000MHz
> revision	: 3.0 (pvr 003c 0300)
> 
> processor	: 1
> cpu		: PPC970FX, altivec supported
> clock		: 2000.000000MHz
> revision	: 3.0 (pvr 003c 0300)
> 
> timebase	: 33333333
> platform	: PowerMac
> model		: PowerMac7,3
> machine		: PowerMac7,3
> motherboard	: PowerMac7,3 MacRISC4 Power Macintosh 
> detected as	: 336 (PowerMac G5)
> pmac flags	: 00000000
> L2 cache	: 512K unified
> pmac-generation	: NewWorld
> 


Reply to: