On Fri, Jul 10, 2026 at 12:20 PM Bryan O'Donoghue <[email protected]> wrote: > > On 10/07/2026 10:04, Loic Poulain wrote: > > + reg = <0x5c42400 0x200>, > > + <0x5c42600 0x200>, > > + <0x5c42800 0x4400>, > > + <0x5c46c00 0x190>, > > + <0x5c46d90 0xa00>; > > + reg-names = "top", "qos", "pipeline", > > + "bus_read", "bus_write"; > > + > > Does this register space cover the entire OPE i.e. where are the CDM > registers ? > > We should map those upfront even if they are not included in your first > drop, we map CDM regs in the IFE even when not used.
My initial thought, especially when we first moved to the CAMSS subnode model, was that CDM would be represented as a separate node and referenced from OPE through a phandle. However, CDM may not be generic enough to justify that level of abstraction/complexity. Therefore, I'll reintroduce the CDM resources here instead. Thanks, Loic > > Also, I looks as though you're missing CDM S1 SID entries in your iommu > definition. > > --- > bod
