Re: PowerMac G5 fans run out of control with kernel 4.17.0-3-powerpc64 but not with 4.16.0-1-powerpc64

[Dennis Clarke <dclarke@blastwave.org>]

On 08/25/2018 06:41 PM, Rick Thomas wrote:
> On Aug 25, 2018, at 7:54 AM, Dennis Clarke <dclarke@blastwave.org> wrote:
>
> > The bucket of things I see back on 4.15.0 are :
> >
> > 4.15.0-2-powerpc64@nix# uname -a
> > Linux nix 4.15.0-2-powerpc64 #1 SMP Debian 4.15.11-1 (2018-03-20) ppc64 GNU/Linux
> > 4.15.0-2-powerpc64@nix# lsmod | grep -i "farm"
> > windfarm_cpufreq_clamp     3085  1
> > windfarm_smu_sensors     7223  1
> > windfarm_smu_controls     7622  8
> > windfarm_pm112         13864  0
> > windfarm_pid            2995  1 windfarm_pm112
> > windfarm_max6690_sensor     3823  1
> > windfarm_lm75_sensor     4317  1
> > windfarm_smu_sat        7208  9 windfarm_pm112
> > windfarm_core          10663  7 windfarm_pm112,windfarm_smu_sensors,windfarm_smu_sat,windfarm_max6690_sensor,windfarm_cpufreq_clamp,windfarm_smu_controls,windfarm_lm75_sensor
>
> Thanks, Dennis…
>
> That’s a slightly different bucket of things than I see on my G5 when it’s running the 4.16 kernel.
> When it’s running the 4.17 kernel *with* the “windfarm_core” in /etc/modules, I get the same bucket as with the 4.16 kernel.
> Without the “windfarm_core” in /etc/modules, the 4.17 kernel shows nothing that meets that grep criterion.
>
> > rbthomas@kmac:~$ uname -a
> > Linux kmac 4.17.0-3-powerpc64 #1 SMP Debian 4.17.17-1 (2018-08-18) ppc64 GNU/Linux
> > rbthomas@kmac:~$ sudo lsmod | grep -i "farm"
> > windfarm_smu_sat        7464  0
> > windfarm_cpufreq_clamp     3213  0
> > windfarm_max6690_sensor     3951  0
> > windfarm_ad7417_sensor     6848  0
> > windfarm_lm75_sensor     4445  0
> > windfarm_fcu_controls    11236  0
> > windfarm_pm72          14394  0
> > windfarm_pid            2995  1 windfarm_pm72
> > windfarm_core          10919  7 windfarm_cpufreq_clamp,windfarm_fcu_controls,windfarm_max6690_sensor,windfarm_smu_sat,windfarm_ad7417_sensor,windfarm_pm72,windfarm_lm75_sensor
>
> I don’t know enough to tell whether the differences between your list and mine are significant.
> All I know is that they both stop the wind-tunnel fans phenomenon.
> I sincerely hope that someone who does know enough to interpret the results can find the time to comment.
>
> Thanks for your help!
> Rick


Well I may have just wasted ten hours on a kernel compile wherein I used 
the config from 4.15.0 and actually removed a number of items but I get 
a initrd that is so obscenely large that there is no way this machine 
will boot :

nix# pwd
/boot
nix# ls -lapb
total 533676
drwxr-xr-x  2 root root      4096 Aug 26 01:26 ./
drwxr-xr-x 23 root root      4096 Oct 22  2017 ../
-rw-r--r--  1 root root   4091837 Mar 20 08:31 System.map-4.15.0-2-powerpc64
-rw-r--r--  1 root root   4103499 Aug 26 01:11 System.map-4.18.5-genunix
-rw-r--r--  1 root root    169433 Mar 20 08:31 config-4.15.0-2-powerpc64
-rw-r--r--  1 root root    159300 Aug 26 01:11 config-4.18.5-genunix
lrwxrwxrwx  1 root root        25 Aug 26 01:24 initrd.img -> 
initrd.img-4.18.5-genunix
-rw-r--r--  1 root root  21791594 Mar 28 20:00 initrd.img-4.15.0-2-powerpc64
-rw-r--r--  1 root root 294498506 Aug 26 01:14 initrd.img-4.18.5-genunix
lrwxrwxrwx  1 root root        25 May 11 02:38 initrd.img.old -> 
initrd.img-4.16.2-genunix
lrwxrwxrwx  1 root root        22 Aug 26 01:11 vmlinux -> 
vmlinux-4.18.5-genunix
-rwxr-xr-x  1 root root  25135392 Mar 20 08:31 vmlinux-4.15.0-2-powerpc64
-rw-r--r--  1 root root 195948280 Aug 26 01:11 vmlinux-4.18.5-genunix
nix#

May as well give it a go .. however that 300MB initrd.img will choke the 
whole process. Really need grub here on this machine but that may not 
help either...

yep https://i.imgur.com/sSjwlEW.jpg

Total waste of time.

I will boot the dvd and rescue that mess and then do a recompile with a
much much smaller initrd situation. Not sure what the limit is or why
the limit exists but I know 300MB is obscene.  Shouldn't be.  However it
is.

Dennis