Bug 4767
| Summary: | RFE: console_blank_hook that can call userspace program | ||
|---|---|---|---|
| Product: | Drivers | Reporter: | Eugene Pavlovsky (heilong) |
| Component: | Console/Framebuffers | Assignee: | Antonino Daplas (adaplas) |
| Status: | REJECTED WILL_NOT_FIX | ||
| Severity: | normal | ||
| Priority: | P2 | ||
| Hardware: | i386 | ||
| OS: | Linux | ||
| Kernel Version: | 2.6.12 | Subsystem: | |
| Regression: | --- | Bisected commit-id: | |
| Attachments: | implemented this functionality in a kernel module | ||
Created attachment 5238 [details]
implemented this functionality in a kernel module
I've actually checked the kernel sources on how this can be done and
implemented the proposed functionality in a kernel module. The module attaches
a console blank hook function that uses kernel's call_usermodehelper() to
execute /sbin/user_vc_blank (defined at compile-time) to blank/unblank the
console. I've written a sample script (which I use on my laptop) that uses
radeontool to turn the LCD backlight off/on. Maybe this functionality should be
moved to the kernel. On the other hand, as it still needs userspace utilities,
maybe it's better off as a module. Anyway, please give it a thought, or at
least made this module available.
Present this to lkml and linux-fbdev-devel list, if you want this functionality. |
I think it'd be very good to have a console_blank_hook handler that would call a userspace program/script/generate hotplug event whatever. Currently, the console can only be blanked using APM, so no options exist for people using ACPI. I've got a Radeon graphics chip on my laptop, and the LCD backlight can be controlled (on/off) using radeontool. If there was a userspace callback interface to console blanking, I would just make a callback script that calls "radeontool light off" on blank and "radeontool light on" on unblank - really easy. I think this userspace console_blank_hook handler is very simple to put into kernel, but I'm not a kernel developer myself, so wouldn't risk sending any patches (that call system("some_script")), because I probably won't make things as they should be in the kernel.