Hi, On Wed, Feb 15, 2017 at 7:27 AM, Javier Martinez Canillas <javier at dowhile0.org> wrote: > Hello Doug, > > On Tue, Feb 14, 2017 at 6:01 PM, Douglas Anderson <dianders at chromium.org> wrote: >> The PMU Cortex M0 on rk3399 is intended to be used for things like >> DDRFreq transitions, suspend/resume, and other things that are the >> purview of ARM Trusted Firmware and not the kernel. As such, the >> kernel shouldn't be messing with the clocks. Add CLK_IGNORE_UNUSED to >> these clocks. >> > > Isn't CLK_IS_CRITICAL a more suitable flag for this case? As I understand it (AKA please correct me if I'm wrong)... Usually CLK_IS_CRITICAL is more suitable than CLK_IGNORE_UNUSED since lots of old code used CLK_IGNORE_UNUSED for critical clocks before CLK_IS_CRITICAL existed. ...but in this case, I don't think it is more suitable. CLK_IS_CRITICAL means that the kernel should be in charge of keeping this clock on at all times. The documentation I see says: #define CLK_IS_CRITICAL BIT(11) /* do not gate, ever */ In our case, as was so eloquently described in our private bug by our firmware engineer: Just tell the kernel to keep its grubby hands off my clocks completely AKA: this isn't a clock that the kernel should touch--it is entirely managed by the firmware. It's OK for the kernel to show it in the clock tree, but otherwise "hands off". -Doug
