[ltp] An "inverse" fan problem

Michael Keyl linux-thinkpad@linux-thinkpad.org
Thu, 26 May 2005 01:17:53 +0200


Zitat von Jan Girlich <vollkorn@freenet.de>: 
 
> Michael Keyl schrieb: 
>  
> > The following C-program can reproduce the effect, although it takes longer 
> to  
> > reach the passive temperature (but the code is not very efficient of 
> course).  
> > I have monitored the temperature as before. 
>  
> I tried to compile the program. But 
> 1. gcc says there is no pentium-m-flag, so I changed it to pentium4,  
> pentium3, pentium, pentiumpro with no effect 
 
Which gcc version do you use? Maybe its too old for pentium-m support. Try 
-mpentiumpro or -mpentium instead. 
 
> 2. Even compiling with using just 
> gcc -o fan_test -lm fan_test.c 
>  
> leads to the same problem: 
>  
> thinkpad ~ # ./fan_test 
> Segmentation fault 
>  
 
Well this is a little bit strange, because I can not reproduce the problem. 
Maybe for your gcc/glibc/kernel the arrays are too big for stack objects. Can 
you try a smaller value for MATDIM (say 10). If this does not work, please 
generate a stack-trace to see where the program crashes: compile with -g flag, 
enable core dumps (ulimit -c unlimited), and then under: gdb fan_test core 
type: bt. 
 
By the way: I have experimented a little bit with the cpu clock, and I have 
figured out that the problem does not occur, if I restrict the clock to 1.4 
Ghz (using /sys/devices/system/cpu/cpu0/cpufreq/scaling_setspeed). So, maybe 
the fan is just underpowered for intensive use of the fp-unit. But it woul be 
interesting to see, if this occurs as well on a 1.5 Ghz machine. 
 
Michael