09:53 <wens> sboyd: IIRC we should not be letting end devices themselves temporarily reparent clocks when upstream PLLs are reconfigured, but instead handle this in the clk drivers with notifiers?

10:39 <marex> I feel like I'm repeating myself, argh

10:39 <marex> sboyd: hi, can we make any progress on https://patchwork.kernel.org/project/linux-clk/patch/20220215084412.8090-2-marex@denx.de/#24780775 resp. https://patchwork.kernel.org/project/linux-clk/patch/20220316123034.13465-1-marex@denx.de/ ?

10:39 <marex> wens: thanks for the reminder

11:04 <arnd> maz: do you have any memory of adding arch/arm/mach-pxa/zeus.c back in 2.6.33? I have a question about PC/104 support on it

11:05 <arnd> my multiplatform series would break ISA I/O space access by moving it to a different virtual address, but I can't tell how it ever worked to start with, as the ports are just mapped to 0xf1000000

11:07 <arnd> ISA drivers would normally expect ports at low numbers. I suppose one could do something like 'modprobe ne io=0xf1000300' and it might work, but I can't tell if that was the intended use

11:09 <arnd> same thing for the other board (viper.c)

11:11 <arnd> It would be trivial to actually fix it so the ports appear at their normal location with my multiplatform series, but that would probably break anything relying on the existing behavior.