https://bugzilla.novell.com/show_bug.cgi?id=387702 User sboyce@blueyonder.co.uk added comment https://bugzilla.novell.com/show_bug.cgi?id=387702#c1 Sid Boyce changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|bnc-team-screening@forge.provo.novell.com |trenn@novell.com --- Comment #1 from Sid Boyce 2008-05-07 06:20:20 MST --- On both laptops # lsmod|grep thermal thermal 29536 2 processor 56388 2 powernow_k8,thermal 1501LCe Boycie:~ # cd /proc/acpi/thermal_zone/THRC Boycie:/proc/acpi/thermal_zone/THRC # o trip_points critical (S5): 98 C passive: 90 C: tc1=2 tc2=5 tsp=300 devices=CPU0 Boycie:/proc/acpi/thermal_zone/THRC # cd ../THRS/ Boycie:/proc/acpi/thermal_zone/THRS # o trip_points critical (S5): 82 C passive: 75 C: tc1=2 tc2=5 tsp=300 devices=CPU0 TravelMate 7520 sepulot:/proc/acpi/thermal_zone/TZS0 # o trip_points critical (S5): 99 C sepulot:/proc/acpi/thermal_zone/TZS1 # o trip_points critical (S5): 110 C As suggested in bug #387207, on the 1501LCe, I have rmmod powernow_k8, thermal and processor and the sensors command still gives 55 degs C. The cpufreq_powersave module is still loaded. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=387702 User sboyce@blueyonder.co.uk added comment https://bugzilla.novell.com/show_bug.cgi?id=387702#c3 --- Comment #3 from Sid Boyce 2008-05-08 13:46:10 MST --- Oops! I left out the data for the 1501LCe which powers off if I boot from 11.0 CD/DVD or build a kernel with powersave set to Performance. When I had the hard drive die, I had to boot and install 10.3, then use zypper dup to get to 110 Alpha and for all the subsequent upgrades zypper dup used. This laptop also running 11.0 Beta2 and sits on a cooler that is separately powered. Building a kernel with "powersave -f". Boycie:/home/boycie # sensors k8temp-pci-00c3 Adapter: PCI adapter Core0 Temp: +76.0°C If left like above, it would have powered off. Boycie:/home/boycie # powersave -l Policy set to powersave Boycie:/home/boycie # sensors k8temp-pci-00c3 Adapter: PCI adapter Core0 Temp: +71.0°C Boycie:/home/boycie # sensors k8temp-pci-00c3 Adapter: PCI adapter Core0 Temp: +69.0°C Boycie:/home/boycie # sensors k8temp-pci-00c3 Adapter: PCI adapter Core0 Temp: +58.0°C Building a kernel with "powersave -l" and it seems stable at 63C. Boycie:~ # sensors k8temp-pci-00c3 Adapter: PCI adapter Core0 Temp: +63.0°C -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=387702 User sboyce@blueyonder.co.uk added comment https://bugzilla.novell.com/show_bug.cgi?id=387702#c5 --- Comment #5 from Sid Boyce 2008-05-08 14:16:15 MST --- Above, "rmmod thermal" done before setting "powersave -f" -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=387702 User trenn@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=387702#c7 Thomas Renninger changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #7 from Thomas Renninger 2008-05-20 10:07:17 MST --- Also see bug #387207. Please reopen if you should still have problems. Hmm, but this happens so early, it cannot have something to do with KDE? If this is not fixed yet, there might be another bug that cpufreq is not activated at installation time. This should already work for AMD cpus (powernow-k8), but maybe it broke for some reason, I double check. It would be great if you could already give the next Beta an installation check. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=387702 User trenn@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=387702#c9 --- Comment #9 from Thomas Renninger 2008-05-21 02:14:10 MST --- You could check here: cat /proc/acpi/thermal_zone/*/* Whether a passive trip point is exceeded (or possibly was) and therefore the machine's frequency gets reduced. Have you rebooted after the upgrade?

I upgraded to 11.0 Beta3 using "zypper dup", now "powersave -f" says "Policy set to Performance", but the frequency doesn't move from 800Mhz. This was during a kernel build. This might be intended. You can monitor temperature: watch -n1 cat /proc/acpi/thermal_zone/*/* If the machine during high load (kernel compile) gets too hot the frequency will be reduced. You also should clean up your fan slots (vacuum cleaner, sometimes you even have to open the machine).

After some time, when the build finished and the machine cools down again, you can reach max_freq again, right? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=387702 User sboyce@blueyonder.co.uk added comment https://bugzilla.novell.com/show_bug.cgi?id=387702#c11 --- Comment #11 from Sid Boyce 2008-05-21 04:40:03 MST --- Laptop has been idle overnight, "powersave -f" doesn't cause speed to increase to 1800 MHz. There is not a trace of dust in the vents and previously when it's powered off, the exhaust temp (felt by hand) is less than on the 7520. It's possible the heatsink to CPU thermal contact isn't what it should be, but I haven't been able to get inside to check it. Boycie:~ # cat /proc/acpi/thermal_zone/*/* 0 - Active; 1 - Passive <polling disabled> state: passive temperature: 95 C critical (S5): 98 C passive: 90 C: tc1=2 tc2=5 tsp=300 devices=CPU0 0 - Active; 1 - Passive <polling disabled> state: passive temperature: 80 C critical (S5): 82 C passive: 75 C: tc1=2 tc2=5 tsp=300 devices=CPU0 Boycie:~ # sensors k8temp-pci-00c3 Adapter: PCI adapter Core0 Temp: +68.0°C -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

Looks as though it has built-in self protection as it has now switched back to 800Mhz and 69 C without using the powersave command. It is not self protection. Or at least not BIOS controlled. At

https://bugzilla.novell.com/show_bug.cgi?id=387702 User trenn@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=387702#c13 --- Comment #13 from Thomas Renninger 2008-05-23 05:29:55 MST --- passive: 90 C degrees the OS will start to throttle the CPU. See /proc/acpi/thermal_zone/*/trip_points Hmm, you do not need sensor tools for watching the temperature. In rare cases sensors and acpi readings could even intefere. When testing, better remove the sensor module. Then do: watch -n1 cat /proc/acpi/thermal_zone/*/{temperature,trip_points,state} (or similar) In another console window (best on the same screen) do: watch -n1 cat /sys/devices/system/cpu/cpu*/cpufreq/scaling_{cur_freq,max_freq} Use dynamic freq control: powersave -A In yet another console start some processes (best as much as cpu cores) e.g.: cat /dev/zero >/dev/null & -> This will produce some heat... You will now see, as soon as the ACPI thermal temperature will exceed 90 degree, max_freq will be limited. After a while, when the CPU cooled down it will get unlimited for some time until 90 degree is exceeded again. (Above /sys and /proc paths might not be 100% correct, the files all exist, but I've just written it down from my mind, there might be typo) -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=387702 User trenn@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=387702#c15 Thomas Renninger changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |hmacht@novell.com --- Comment #15 from Thomas Renninger 2008-05-23 09:47:26 MDT ---

Sometimes I get "good boots" and sometimes I get "bad boots".

Miguel, better stay on your original report, the machines are too different. I think I will connect all ThinkPad 11.0 temp bugs somehow and then try to ask mobile team for a reference machine. Or better they should look at it, I am off for holidays for some days. base stations..., something comes to my mind. We load the dock driver quite early now, maybe the ACPI functions there on the ThinkPad trigger something. Removing it may not be that easy, AFAIK ata drivers depend on it (that's why it is loaded that early). Holger how can the dock/bay driver be disabled? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=387702 User hmacht@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=387702#c17 --- Comment #17 from Holger Macht 2008-05-23 10:03:37 MDT --- The only thing the dock driver does is executing the _DCK method. If at all, when the system was turned on with the docking station attached. I doubt this would trigger any problems. Regarding the second issue, libata drivers need to be notified by the dock driver for undock/dock events to be able to care about hotplug devices inside the dock stations. I don't see why this should be hacky, nor can imagine a better solution. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=387702 User trenn@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=387702#c19 --- Comment #19 from Thomas Renninger 2008-05-23 15:07:13 MDT --- Yes, I thought a while about it, but I also have no idea how to do it nicer. Theoretically if in the ACPI base there is an interface where several drivers can register to get notifications for a device and notify handlers are installed in the ACPI core and then dispatched... I better stop here this gets utopistic. Maybe via a global dock notifier chain that is automatically set up in ACPI subsystem? Everybody can register for it, dock triggers the notify, could this work? Probably it is not worth it, I just need to get used to that there is the dock driver added to initrd. Was no offence, on a first glance it just looked a bit ugly. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.