Andy Green wrote: > It was to have been unavoidable on GTA03, the s3c6410 has a broken pin > allocation for card detect hardware pin, you had to nudge the stack to > tell it that a connect or disconnect action had occurred.
Yuck. > But "that issue has gone away now". Gee, so there is a bright side to it :-) > You fancy adding this? I guess it should do whatever it is the ioctl does. I wonder what the rfkill semantics should be when it comes to associating and other ioctls. The two ioctls differ slightly in what happens inside the kernel: wmiconfig --wlan: - disable: de-associate and stop scanning - enable: start scanning and associate SIWTXPOW: - disable: de-associate and stop scanning - enable: start scanning but don't associate wmiconfig --wlan also sets a more global state, so most operations (ioctls, mainly) are refused (EIO) while WLAN is disabled. If rfkill is to act just like a switch that cuts power to the RF amp, then we'd actually a combination of the two. - Werner
