Changing the cpu speed (up or down) on my Time Platina (Arima built laptop) amd64 laptop causes it to lock up completely. Kernel is 2.6.5-r1 gentoo-dev-sources. The same problem is evident in both x86 and x86_64 versions. Fedora Core 2 (32bit) works just fine with changing cpu speed. With its default kernel (2.6.5-1.358) recompiled to include powernow-k8. Reproducible: Always Steps to Reproduce: 1. cd /sys/devices/system/cpu/cpu0/cpufreq 2. echo 800000 > scaling_setspeed (or use something like powernowd or even change the governor) Actual Results: The system locks and the cpu fan goes into overdrive. Expected Results: Change the cpu speed to 800mhz
Despite spending several minutes changing the cpu speed up and down and up and down on Fedora before filling this bug report without any crashes I have just had Fedora crash when changing the cpu speed.
From using Fedora which crashes considerably less frequently than gentoo (why I have no idea) I have discovered changing the speed creates an error. Now this doesn't appear to be a problem at first and you can change the cpu speed without any problems several times despite the error. However eventually this error gets caught in some kind of loop and grabs all the cpu processes and crashes out the laptop Error info:- ------------------------- Debug: sleeping function called from invalid context at include/linux/rwsem.h:43 in_atomic():1, irqs_disabled:0 Call Trace: [<02117ea6>] __might_sleep+0xa5/0xaf [<02272524>] cpufreq_notify_transition+0x2e/0x141 [<0210edc5>] transition_frequency+0x8d/0exb [<0210eee3>] powernowk8_target+0xc0/0x105 [<02271fab>] __cpufreq_driver_target+0xa/0xc [<0227295d>] cpufreq_set+0x66/0x7d [<022729ef>] store_speed+0x31/0x39 [<022729be>] store_speed+0x0/0x39 [<02271b54>] store+0x2e/0x3e [<02174ecb>] flush_write_buffer+0x1d/0x22 [<02174ef2>] sysfs_write_file+0x22/0x35 [<021469db>] vfs_write+0xb8/0xe4 [<02146a57>] sys_write+0x2c/0x42 ----------------------------- I can only get this info from my Fedora install because it only crashes sometimes. On gentoo it crashes all the time and the text kinda cycles up the screen to fast to read. However I suspect the same thing is behind the crash.
<quote>The same problem is evident in both x86 and x86_64 versions.</quote> => Reassigning to x86-kernel@g.o
Does this still happen on the 2.6.7-r2 gentoo-dev-sources release?
Oh, and if it is still a bug with 2.6.7, this is a upstream issue. If so, please file a bug at bugzilla.kernel.org.