On Tue, 2009-07-28 at 10:35 +0200, Steve Langasek wrote: > FWIW, I'm still unclear why a special tool is needed to manage killswitches > now, since it used to be possible to set these directly via the interfaces > under /sys - the interfaces appear to still be there, but they no longer > accept changing the values, requiring access via a control device under > /dev instead. Surely this is a regression from the perspective of the > kernel design? That has been changed, it accepts changing the value in /sys again. However, the value in sysfs can only represent three states: * radio on * hardware blocked * software blocked it cannot distinguish between "hard-blocked & soft-blocked" and "hard-blocked & !soft-blocked". Additionally, uevents are hard to manage for notifications, and /dev/rfkill is also required for moving the input button handling from the kernel into userspace. So the rfkill tool isn't really required, but it's very useful for debugging. Try telling a user to look around in /sys to find all the rfkill information that 'rfkill list' prints out [1]. johannes [1] even if it was possible (see above hard vs. soft blocked)
Attachment:
signature.asc
Description: This is a digitally signed message part