> Subject: Re: [PATCH v7 1/3] dt-bindings: soc: imx93-media-blk-ctrl:
> Add PDFC subnode to schema and example
> 
> > We discussed the ranges/reg usage internally as well with the
> > following
> > outcome:
> >
> > - Situation
> >
> >   With i.MX8M SoCs NXP introduced MIX domains. Each domain has
> a so
> >   called BLK-CTRL IP. There is no common register layout for the
> >   BLK-CTRL IPs. In addition to this, the register fields within one
> >   register may not related to the same IP. Please see my below
> example:
> >
> >   The DISP_MUX register configures the DPI output routed to physical
> >   SoC pads as well as the internal MIPI-DSI DPI behavior. This PDFC
> >   bridge binding is only interested in the first part, not the 2nd.
> >
> >   In other words, the BLK-CTRL IP can be seen as a bunch of loose
> >   register fields.
> 
> Why do hardware engineers keep on insisting on doing this. And why
> don't software engineers when asked to review the proposed registers
> push back and say No?
> 
> I know this is out of you hands, but it is something we should
> encourage software engineers at SoC vendors to do. Otherwise we will
> have to continually live with such a mess.

This is kind of system architecture. There are several mixes in the SoC,
each mix is a group of IPs(in-house designed or 3rd party). The blk-ctrl
module is to do various settings that could be done in each IP.

I agree that BLK-CTRL is not good for software usage, and we need to
discuss with architectures for future improvements.

Thanks,
Peng.

> 
>      Andrew

Reply via email to