It doesn't matter what value I use, it is treated as if unset (i.e. "0"). I guess if I had a module to reload, it would probably set the correct value. But the kernels on ARM hardware often have such drivers build in the kernel directly. I want to stress that this is important on ARM systems. With the normal polling rate of 100Hz, Youtube may use up to 60% additional CPU in Firefox on a 2017 TV box, making either 1080p videos stutter or throttle down to 144p all the time. With a gaming mouse that defaults to 1000Hz, it may totally max out the CPU even on 480p. On a TV box, people rather use low-accuracy mini-keyboard trackpads than normal mice. Those can be set to 20Hz and as low as 10Hz without any difference in usability. As a person might use both the remote mini-keyboard from the couch to watch TV and also a normal set of mouse/keyboard when sitting in front of the TV and using the TV box for office applications or browsing, it would be best if mousepoll could be set per device instead. usbhid.mousepoll=50 > cat /sys/module/usbhid/parameters/mousepoll 50 > ./evhz HID 0e8f:2517 Mouse: Latest 62Hz, Average 91Hz HID 0e8f:2517 Mouse: Latest 125Hz, Average 91Hz HID 0e8f:2517 Mouse: Latest 62Hz, Average 91Hz HID 0e8f:2517 Mouse: Latest 125Hz, Average 91Hz ... Razer Razer DeathAdder: Latest 1000Hz, Average 921Hz Razer Razer DeathAdder: Latest 1000Hz, Average 921Hz Razer Razer DeathAdder: Latest 1000Hz, Average 921Hz Razer Razer DeathAdder: Latest 1000Hz, Average 921Hz It is working on my 5.10.9-arch1-1 x86 PC with module reload, unplugging and plugging the cable back in.
On Sun, Jan 24, 2021 at 04:35:13AM +0000, bugzilla-daemon@bugzilla.kernel.org wrote: > https://bugzilla.kernel.org/show_bug.cgi?id=211325 > > Bug ID: 211325 > Summary: usbhid.mousepoll kernel cmd option not working > Product: Drivers > Version: 2.5 > Kernel Version: 5.11-rc3 > Hardware: ARM > OS: Linux > Tree: Mainline > Status: NEW > Severity: low > Priority: P1 > Component: USB > Assignee: drivers_usb@kernel-bugs.kernel.org > Reporter: mathias.steiger@gmail.com > Regression: No > > It doesn't matter what value I use, it is treated as if unset (i.e. "0"). Please email the linux-input@vger.kernel.org list about this.