hid-core and cpufreq conflicts
Hello,
I am having an interesting conflict between USB and the CPUfreq driver for P4 mobile in my laptop... (Compaq Evo N610c)
When I have both enabled, and the CPUfreq set to use the "userspace" governor, hid-core crashed with the following error every time I click the mouse buttom more than once in a second. (so double clicking is a pain) I have not had any other USB actions cause this crash, such as using my USB connected wireless card.
(from dmesg)drivers/usb/input/hid-core.c: can't resubmit intr, 0000:02:0e.2-4.1/input0, status -32
After this error I have to rmmod ehci_hcd && service usb restart to make anything usb work again.
I have also noticed that the frequency of the processor will not scale below 1196mhz (processor is a 2.0ghz P4 mobile) when ehci_hcd is loaded, but it will scale all the way down to 130mhz when this module is not loaded...
Any ideas on what may be causing the problems?
Thanks,
Jeff Hagen
[email protected]
Please CC me in any replies as I am not a member of the USB-users mailinglist.
Hi,
Quoted 3 lines · Jeff Hagen
I have also noticed that the frequency of the processor will not scale below 1196mhz (processor is a 2.0ghz P4 mobile) when
ehci_hcdis loaded, but it will scale all the way down to 130mhz when this module is not loaded...
If it's a P4 mobile, you should use either the acpi module or the speedstep-ich module for frequency scaling, not p4-clockmod for frequency throttling. Scaling down to 130Mhz sound strange anyway...
Quoted 1 line
Any ideas on what may be causing the problems?
p4clockmod may interfere with latencies as the CPU is stopped for short periods of time. Maybe I should change arch/i386/kernel/cpu/cpufreq/Kconfig to
config P4CLOCKMOD_ENABLE
bool "p4-clockmod"
help
Are you really sure you want to compile the p4-clockmod driver which in most cases does really nothing besides causing pain?
config P4CLOCKMOD_REALLY_ENABLE
bool "really p4-clockmod?"
depends on P4CLOCKMOD_ENABLE
help
Have you really thought about it?
Dominik