On Wed, Jul 24, 2024 at 01:55:41PM +0200, Krzysztof Kozlowski wrote: > On 24/07/2024 13:41, Varadarajan Narayanan wrote: > > On Wed, Jul 24, 2024 at 08:27:03AM +0200, Krzysztof Kozlowski wrote: > >> On 23/07/2024 11:03, Varadarajan Narayanan wrote: > >>> USB uses icc-clk framework to enable the NoC interface clock. > >>> Hence the 'iface' clock is removed from the list of clocks. > >>> Update the clock-names list accordingly. > >> > >> But the clock is still there and is still used by this block. This looks > >> like adjusting hardware per Linux implementation. > >> > >> Why suddenly this clock was removed from this hardware? > > > > This clock per se is not used by the USB block. It is needed to > > enable the path for CPU to reach the USB block (and vice versa). > > Hence, we were adviced to use the ICC framework to enable this > > clock and not the clocks/clock-names DT entries. > > > > Please refer to [1] where similar clocks for IPQ9574 were NAK'ed. > > So the original submission was not correct? Unlike MSM SoC, IPQ SoC doesn't use RPM to aggregate bandwidth requests and scale the NoC frequency. The NoCs are turned on and set to a specific frequency at boot time and that is used for the lifetime of the system. Hence interconnect was not considered and this submission was accepted. The same approach was used for PCIe and at that point the consensus was to move to interconnect. Hence implemented the ICC driver and updating the existing USB driver to use the ICC driver. > You really need to stop sending DTS based on current driver support and > focus on proper hardware description. > > Such things pop up from time to time for Qualcomm and I don't see much > of improvement. And we do not talk about some ancient code, predating > guidelines, but IPQ5332 upstreamed ~1 year ago. We are trying, but falling short. Hopefully we meet the expectations soon. Thanks Varada
