On Mon, 14 Aug 2023 12:10:25 -0600 Sam Edwards <cfswo...@gmail.com> wrote:
> Hi Andre, > > On 8/14/23 10:37, Andre Przywara wrote: > > So I think we can get rid of this: > > - GEN_H6 never compiles this code here, as both H6 and H616 are arm64. > > Easy! > > > - We can define SUNXI_PRCM_BASE for NCAT2, I believe Samuel once > > mentioned that the D1/T113 does have such a block, actually. > > Will you be taking care of this in v2 of your T113s series, or should I > be adding it (in which case I'll need to know the location of the block)? Yes, I will add this to the header file, either defined as 0, or to its actual address. > > - The non-existing cpu_pwr_clamp member should go away when you switch to > > a BASE_ADDR + REG_OFFSET approach, I think. > > Less easy, but still can do. > > > Shouldn't that be the opposite? In the existing code, sun6i and H3 DO > > program the clamp (see the "-" section above). > > And sun7i and R40, as well. Yes, but you handle both above explicitly. > It appears I simply read the #if > defined(...) mess backwards. I'll fix that for v2. As a bonus, this > lends itself to a rather nice refactoring of sunxi_cpu_set_power() where > I can have the if block only determine the pwroff/clamp addresses, and > have a single tail-call to sunxi_power_switch() at the bottom. Since the > latter function is so simple, I may as well just inline it into > sunxi_cpu_set_power() (which I suspect might be more readable). Yes, any further simplification is welcome, and probably somewhat rewarding in this case ;-) Cheers, Andre
