https://bugzilla.novell.com/show_bug.cgi?id=351915 User alpha096@virginbroadband.com.au added comment https://bugzilla.novell.com/show_bug.cgi?id=351915#c2 Scott Couston changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW Info Provider|alpha096@virginbroadband.com.au | --- Comment #2 from Scott Couston 2008-01-11 15:59:17 MST --- Created an attachment (id=190305) --> (https://bugzilla.novell.com/attachment.cgi?id=190305) Hardware info as requested Note 1. you command "output of 'hwhinfo'" yields bad command. I have assumed 'hwinfo' in lieu. Output of command 'hwinfo' is attached. Note 2. This PC is my File Server/Print Server/Installation Server and it is summer down under. The PC was shut down by BIOS due overheat 1 day this week. The unit has 2 x 120mmCase Fans and 1 CPU heat sink and cooling fan. Note 3. If I have interpreted your command incorrectly please advise see Not 1 -- 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=351915 User trenn@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=351915#c3 Thomas Renninger changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WORKSFORME --- Comment #3 from Thomas Renninger 2008-02-10 14:15:11 MST --- throttling != CPU frequency scaling CPU frequency scaling is to reduce power consumption. Throttling is more to prevent overheating (fans are a better approach for that...) I doubt that your Celeron CPU is capable of "Enhanced Speedstep Technology" EST, the Intel implementation of CPU freq. Please read your CPU's documentation, I am pretty sure your processor cannot do it, if yes, please point to the relevant docu. Throttling is not very interesting (far away from critical..., maybe minor or enhancement), but it should work through /proc/acpi/processor/*/throttling interface if supported. -- 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=351915 User trenn@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=351915#c5 Thomas Renninger changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |NEEDINFO Info Provider| |alpha096@virginbroadband.com.au Summary|NO CPU Throttling applied to I386 Intel Celeron |I386 Intel Celeron 2.4Mhz shows critical thermal |2.4Mhz |shutdowns --- Comment #5 from Thomas Renninger 2008-02-10 15:28:25 MST --- throttling is rather unimportant, thermal shutdowns not -> adjusting title. Please check that you do not have such problems: https://bugzilla.novell.com/show_bug.cgi?id=359307#c2 Have you modified fans (CPU, chassis, ...) since you bought the machine? Please attach /var/log/messages (possibly zipped if too large). Is the BIOS throttling/cpufreq setting set to software or hardware now? Please provide /proc/acpi/thermal_zone/*/trip_points You should be able to monitor temperature with: watch -n1 cat /proc/acpi/thermal_zone/*/temperature Do: cat /dev/zero >/dev/null & two times and watch the temperature in another shell for about 5-10 minutes. How high does the temperature grow? -- 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=351915 User trenn@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=351915#c7 --- Comment #7 from Thomas Renninger 2008-02-11 04:13:05 MST --- Can you also provide other info from comment #5, pls: -------- Please provide /proc/acpi/thermal_zone/*/trip_points You should be able to monitor temperature with: watch -n1 cat /proc/acpi/thermal_zone/*/temperature Do: cat /dev/zero >/dev/null & two times and watch the temperature in another shell for about 5-10 minutes. How high does the temperature grow? It may be that your BIOS does not export a thermal zone, then /proc/acpi/thermal_zone is empty, is that the case? I expect one of these two things are happening: 1) Interference with other thermal drivers and (possibly, but could be something else) ACPI. E.g. smartd or lmsensors (Do you use lmsensor modules? This could be the reason for your rare, unexpected shutdowns) read the temperature sensor while ACPI is doing it. This normally results in totally insane values. E.g. critical shutdown of 3000 C Therefore I also wanted to have /var/log/messages output. If you do a: grep -i critical /var/log/messages You should find the line when the machine shut down. There also should be in brackets the temperature value. I need this value. There are patches in the very latest kernel which might detect this interference problems, I can point you to them if this could be the problem. If you use lmsensors, this might be the causem, but we could verify with the patches then. The phenomenon if you hit this, is that you get irregular shutdowns (e.g. also if the machine is cool), every day/week/month/random. 2) Fan/power management might be messed up or the fans are not strong enough In this case the shut downs should easily be reproducable for you. Please monitor the temperature (as stated above) and give your CPU something to do (maybe also the graphics card, glxgears is a nice tool for that). The temperature should grow constantly and (nearly?) reach the critical temperature. -- 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=351915 User alpha096@virginbroadband.com.au added comment https://bugzilla.novell.com/show_bug.cgi?id=351915#c9 Scott Couston changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |RESOLVED Info Provider|alpha096@virginbroadband.com.au | Resolution| |REMIND --- Comment #9 from Scott Couston 2008-02-16 21:55:12 MST --- PC returned after additional HDD added, Load New O/S and run under initial fault environment - Waiting for 1. Sufficient data collection as requested 2. Ambient temp rises to sufficient initial test/bug report and output log can be gathered. Change status - Remind Reply on cooling management - I have another 2 x P4 HT 3.2 I386 with same case and cooling ability - They cannot demonstrate insufficient lack of cooling ability. Check physical fans operation - Fans demonstrate ability to dynamically adjust air flow to comply with purpose, design,and operation. -- 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=351915 User alpha096@virginbroadband.com.au added comment https://bugzilla.novell.com/show_bug.cgi?id=351915#c11 Scott Couston changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|REMIND | --- Comment #11 from Scott Couston 2008-02-25 00:32:40 MST --- Please advise which files or packages are required from Software Management>Search to enable above command "You should be able to monitor temperature with: watch -n1 cat /proc/acpi/thermal_zone/*/temperature Do: cat /dev/zero >/dev/null & " The directory /proc/acpi/ does contain the directory */thermal_zone/ and ALL other subdirectories of /proc/acpi are empty of files. Please check syntax I think switch 'l' is not an available switch from usage of the command 'watch' Please confirm command is executable via a normal user. I have tried both root and normal user - seems that root is not required???????? -- 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=351915 User trenn@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=351915#c12 Thomas Renninger changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |REOPENED Info Provider|trenn@novell.com | --- Comment #12 from Thomas Renninger 2008-02-25 01:20:01 MST ---

I think switch 'l' is not an available switch It is a one. watch -n1 ".." will execute the following command every second and show its output.

If /proc/acpi/thermal_zone/ is emtpy: That means your machine's BIOS does not export a Thermal Zone device via ACPI (and thermal driver is loaded). How do think that the machine shutdowns were temperature related? If the thermal driver is involved, it will check the temperature and shut down the machine gracefully if the critical thermal trip point is exceeded. You will see a message in /var/log/messages: Critical temperature reached, shutting down (or similar, the current temperature is also stated there). If it is not the thermal driver (in your case cannot be if thermal_zone is empty), you cannot see this message. If your machine hangs/shut downs are really temperature related it will simply turn off, but this can also come from other problems. E.g. I would first try a BIOS update and a memory test if you see hard switch offs. -- 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=351915 User alpha096@virginbroadband.com.au added comment https://bugzilla.novell.com/show_bug.cgi?id=351915#c14 Scott Couston changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |RESOLVED Info Provider|alpha096@virginbroadband.com.au | Resolution| |INVALID --- Comment #14 from Scott Couston 2008-02-25 14:52:33 MST --- BIOS is current and monitor temp of CPU and RAM - It was the BIOS that turned off this PC due to BIOS monitoring reached critical temperature - Hence the issue that the CPU was not throttled to prevent turn off (sorry error in terminology) As there is NO ACPI driver installed I cannot run the above. As you pointed out if the ACPI directories are empty the PC BIOS cannot export temperature so there is no resolution nor fault on your half - closed -I will just have to lower the continual CPU load by moving an application monitor that is very hungry on resources. -- 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=351915 User alpha096@virginbroadband.com.au added comment https://bugzilla.novell.com/show_bug.cgi?id=351915#c16 --- Comment #16 from Scott Couston 2008-02-26 20:48:35 MST --- Before creating this possible kernel issue I noted that another PC on the Network which has the same Motherboard/BIOS Rev...etc but has a Intel P4 HT 3.2 CPU. 10.3 does throttle the CPU when required and can report on temps. This unit has the same case fan as reported, however this unit has only 1x 120mm dynamic case fan the above unit has 2 x 120mm dynamic case fans. The BIOS also reports the RPM of the Major Case Fan and came with a MS Windows CD that permits setting of user defined temps/RPM (dynamic) to send pop-ups to the user whilst using any MS Windows XP APP. There are also default temp/RPM values that will also cause pop-ups to occur. From my memory I think it also permitted the user to over-ride the Dynamic RPM of Primary Case Fan to a value above, but not limited to the value imposed by the user. This app sat at in the MS Windows XP Pro system Tray. All motherboards in PC's on the Network are made by ASUS and I have lobbied them to start to offer all goodies that come with the motherboards like 'Internet Dynamic Auto Update of BIOS' and above type utilities; to come in an Linux .RPM format. Keep finger crossed one day they will ;-) As an aside the P4 HT 3.2MHz CPU is almost equal to the AMD Athlon 64 X2 Dual Core Processor 4800+ 2.6MHz as far as performance with OpenSuse 10.3 and it even boots and loads and shuts down faster that the X_64 CPU. I am really very very disappointed in the X_64 CPU performance and was expecting a lot more that the P4 3.2 HT provides. -- 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=351915 Scott Couston changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |NEEDINFO Info Provider| |trenn@novell.com -- 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=351915 User alpha096@virginbroadband.com.au added comment https://bugzilla.novell.com/show_bug.cgi?id=351915#c19 --- Comment #19 from Scott Couston 2008-02-28 16:02:50 MST --- Regardless of where or what manner CPU temp is monitored and throttled this is an example of the other P4 HT 3.2 where NO acpi files are present. The bug report was on the same motherboard,same BIOS etc but with an Intel Celeron CPU as title. - THAT LOG IS EMPTY and there is also no ACPI interface. As the processor is no longer in production I can understand pushing for a kernel change that is not only downward compatiable but not in production. P4 HT 3.2 Log, same BIOS, motherboard, NO ACPI interface Dec 20 15:15:37 linux kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 1510) Dec 20 15:15:38 linux kernel: CPU1: Temperature/speed normal Dec 21 01:15:38 linux SuSEfirewall2: SuSEfirewall2 not active Dec 20 15:34:17 linux kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 2005) Dec 20 15:34:17 linux kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 2018) Dec 20 15:41:41 linux kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 2032) Dec 20 15:41:41 linux kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 2045) Dec 20 15:46:58 linux kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 4393) Dec 20 15:46:58 linux kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 4380) Dec 20 16:00:02 linux kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 6000) Dec 20 16:00:02 linux kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 5987) Dec 20 16:13:24 linux kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 7327) Dec 20 16:13:24 linux kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 7340) Dec 20 16:18:40 linux kernel: JBD: barrier-based sync failed on dm-0 - disabling barriers Dec 20 16:18:41 linux kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 11233) Dec 20 16:18:41 linux kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 11246) Dec 20 16:23:51 linux kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 13265) Dec 20 16:23:51 linux kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 13252) Dec 20 16:33:52 linux kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 15806) Dec 20 16:33:52 linux kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 15819) Dec 20 16:38:52 linux kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 19890) Dec 20 16:38:52 linux kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 19877) Dec 20 16:47:44 linux kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 19918) Dec 20 16:47:44 linux kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 19905) Dec 20 16:52:55 linux kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 24636) Dec 20 16:52:55 linux kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 24649) Dec 20 16:57:55 linux kernel: CPU1: Temperature/speed normal Dec 20 16:57:55 linux kernel: CPU0: Temperature/speed normal Dec 20 16:57:55 linux kernel: CPU0: Temperature/speed normal Dec 20 17:03:19 linux kernel: CPU1: Temperature/speed normal Dec 20 17:08:22 linux kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 30360) Dec 20 17:08:22 linux kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 30347) Dec 20 17:37:16 linux kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 31055) Dec 20 17:37:16 linux kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 31068) Dec 20 17:56:10 linux kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 35992) Dec 20 17:56:10 linux kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 35979) -- 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=351915 User alpha096@virginbroadband.com.au added comment https://bugzilla.novell.com/show_bug.cgi?id=351915#c21 --- Comment #21 from Scott Couston 2008-03-01 15:47:27 MST --- re #18 - There is NO HW cooling issue - The Kernel will not throttle as per bug title. The above log in #19 shows clearly how the Kernel does throttle the CPU when its temp is above threshold. The log on the Celeron CPU does not mention any temperature issues and also indicates the Kernel has not throttled the CPU. The variant in this situation is simply the ambient temperature, as it is summer here in OZ this PC will and does shut itself down via a BIOS function. SO we are back to the title of the bug. QUOTE(sic) BIOS is monitoring the temperature. If it detects a critical temp situation it can throw a kind of CRIT_TEMP event to the kernel through some kind of machine check execption interface, so that the user can be informed. This is implemented here: arch/i386/kernel/cpu/mcheck/therm_throt.c This is an Intel specific (already deprecated?) mechanism UNQUOTE This mechanism is demonstrated with the P4 HT 3.2 CPU log and shows the kernel throttling the CPU frequency when required. As the same log for the Celeron CPU PC with same BIOS/Motherboard, has no noted information on CPU temp and throttling so we are back at the title of the BUG. I understand that authorising a kernel change to support ANY legacy device is difficult enough to rationalise and authorise by QA Manager. As the fix would require a change to the Kernel for the sole reason of providing legacy hardware support would be a hard call and if was back working in development and had to authorise such a change; I would not authorise it also for many reasons too long to discuss here. This bug is already verbose and I am happy with the resolution. Its time I upgraded the motherboard/CPU as the current system needs 24/7 uptime as it is a file server also running SQL Data Security Management software which is gathers from many client via VPN. -- 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.