25863549be
This feature was a useless noop based on mistaken API understanding. The idea seems to have been that k_busy_wait() included guards to ensure "clock_always_on" was true duing the loop, presumably because the original author was afraid that "turning the clock off" would affect the operation of k_cycle_get_32(). Then later someone came around and "optimized" this for Quark SE, where the cycle counter is the RTC and unrelated to the timer driver used by the clock_always_on feature. (Except even there it presumably should have been done at the SoC level and not just in the C1000 devboard -- note that Arduino 101 never would have gotten this). But it was all a mistake: "clock_always_on" has nothing to do with en/disabling the system cycle timer (which never happens when the system is active, that's a feature of idle), it's a control over the delivery of timer interrupts. And needless to say we don't care about timer interrupts when we're spinning on a cycle counter. Yank the whole mess. Signed-off-by: Andy Ross <andrew.j.ross@intel.com> |
||
|---|---|---|
| .. | ||
| arduino_101 | ||
| common/scripts | ||
| galileo | ||
| minnowboard | ||
| qemu_x86 | ||
| quark_d2000_crb | ||
| quark_se_c1000_devboard | ||
| tinytile | ||
| up_squared | ||
| x86_jailhouse | ||
| index.rst | ||