Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Bug#603981: initramfs-tools: Load PowerMac G5 thermal modules

6 views
Skip to first unread message

Milan Kupcevic

unread,
Nov 18, 2010, 9:20:01 PM11/18/10
to
Package: initramfs-tools
Version: 0.98.5
Severity: important
Tags: patch

On iMac and PowerMac G5 machines, about a minute after boot, fans run at
full speed producing jet engine noise. Thermal modules do not get loaded
as they were in Lenny.

initramfs-tools-load-G5-thermals.patch
signature.asc

Milan Kupcevic

unread,
Nov 18, 2010, 10:40:01 PM11/18/10
to
The first patch did not include PowerMac9,1. Corrected patch is attached
to this message.

Thanks,

Milan

initramfs-tools-select-G5-thermals.patch
signature.asc

maximilian attems

unread,
Nov 19, 2010, 4:20:02 AM11/19/10
to
On Thu, Nov 18, 2010 at 10:09:59PM -0500, Milan Kupcevic wrote:
> The first patch did not include PowerMac9,1. Corrected patch is attached
> to this message.

oh fun, thought that linux-2.6 was fixed to autoload those modules.

> diff -Nru ./hooks/thermal ../initramfs-tools-0.98.5/hooks/thermal
> --- ./hooks/thermal 2010-09-23 14:43:51.000000000 -0400
> +++ ../initramfs-tools-0.98.5/hooks/thermal 2010-11-18 21:54:21.000000000 -0500
> @@ -22,19 +22,36 @@
> case "$DPKG_ARCH" in
> # copy the right modules
> powerpc|ppc64)
> - if [ -e /sys/bus/ps3_system_bus/ ]; then
> - exit 0
> - fi
why do you remove the ps3 check??

> - manual_add_modules therm_pm72
> +
> + # Only PowerMac G5 machines need these modules
> +
> + MODEL="`grep model /proc/cpuinfo 2>/dev/null`"; MODEL="${MODEL##*: }"
please 2 lines and a check that /proc/cpuinfo is readable
(think of chroots)
> +
> + case "$MODEL" in
> + RackMac3,1|PowerMac7,2|PowerMac7,3)
> + force_load therm_pm72
> + ;;
> + PowerMac8,1|PowerMac8,2)
> + force_load windfarm_pm81
> + ;;
> + PowerMac9,1)
> + force_load windfarm_pm91
> + ;;
> + PowerMac11,2)
> + force_load windfarm_pm112
> + ;;
> + PowerMac12,1)
> + force_load windfarm_pm121
> + ;;
> + *)
> + exit 0
hmm why this exit, seems bad for any other box?
> + ;;
> + esac
> manual_add_modules windfarm_core
> manual_add_modules windfarm_cpufreq_clamp
> manual_add_modules windfarm_lm75_sensor
> manual_add_modules windfarm_max6690_sensor
> manual_add_modules windfarm_pid
> - manual_add_modules windfarm_pm121
> - manual_add_modules windfarm_pm112
> - manual_add_modules windfarm_pm81
> - manual_add_modules windfarm_pm91
> manual_add_modules windfarm_smu_controls
> manual_add_modules windfarm_smu_sat
> manual_add_modules windfarm_smu_sensors

otherwise this looks like a good workaround for this kernel bug.
please repost fixed patch.


--
To UNSUBSCRIBE, email to debian-bugs-...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listm...@lists.debian.org

Milan Kupcevic

unread,
Nov 20, 2010, 1:30:01 AM11/20/10
to
On 11/19/2010 04:09 AM, maximilian attems wrote:

> oh fun, thought that linux-2.6 was fixed to autoload those modules.

Only i2c_powermac gets included/loaded without manual intervention. I
removed it from this hook.


>> - if [ -e /sys/bus/ps3_system_bus/ ]; then
>> - exit 0
>> - fi
> why do you remove the ps3 check??

No need to exit here on PS3 because this patch includes/loads therm_pm72
and windfarm_* modules on G5 Mac machines only.

>
>> - manual_add_modules therm_pm72
>> +
>> + # Only PowerMac G5 machines need these modules
>> +
>> + MODEL="`grep model /proc/cpuinfo 2>/dev/null`"; MODEL="${MODEL##*: }"
> please 2 lines and a check that /proc/cpuinfo is readable

OK. Fixed.

>> +
>> + case "$MODEL" in
>> + RackMac3,1|PowerMac7,2|PowerMac7,3)
>> + force_load therm_pm72
>> + ;;
>> + PowerMac8,1|PowerMac8,2)
>> + force_load windfarm_pm81
>> + ;;
>> + PowerMac9,1)
>> + force_load windfarm_pm91
>> + ;;
>> + PowerMac11,2)
>> + force_load windfarm_pm112
>> + ;;
>> + PowerMac12,1)
>> + force_load windfarm_pm121
>> + ;;
>> + *)
>> + exit 0
> hmm why this exit, seems bad for any other box?

Every G5 Mac model [1] is covered. We can safely exit here because no
other box is using windfarm_* modules.

>> + ;;
>> + esac
>> manual_add_modules windfarm_core
>> manual_add_modules windfarm_cpufreq_clamp
>> manual_add_modules windfarm_lm75_sensor
>> manual_add_modules windfarm_max6690_sensor
>> manual_add_modules windfarm_pid
>> - manual_add_modules windfarm_pm121
>> - manual_add_modules windfarm_pm112
>> - manual_add_modules windfarm_pm81
>> - manual_add_modules windfarm_pm91
>> manual_add_modules windfarm_smu_controls
>> manual_add_modules windfarm_smu_sat
>> manual_add_modules windfarm_smu_sensors
>
> otherwise this looks like a good workaround for this kernel bug.
> please repost fixed patch.

New patch is attached.


[1]
http://www.everymac.com/systems/by_capability/mac-specs-by-machine-model-machine-id.html

initramfs-tools-select-G5-thermals.patch
signature.asc

maximilian attems

unread,
Nov 23, 2010, 5:10:03 AM11/23/10
to
On Sat, 20 Nov 2010, Milan Kupcevic wrote:

>
> New patch is attached.

thank you very much, merged for review in maks/hook_thermal
http://git.debian.org/?p=kernel/initramfs-tools.git;a=summary

--
maks

maximilian attems

unread,
Nov 23, 2010, 6:00:01 AM11/23/10
to
On Tue, Nov 23, 2010 at 11:30:21AM +0100, Mattia Tristo wrote:
> I have installed Debian 5.0 PPC testing on my PowerMac 7,3
> tomorrow in the afternoon i will test the patch
> Thanks

how *often* do I have to repeat that, no private mails!

if you have to say something in public, do so.
No private support, thank you.

0 new messages