On Tue, Mar 30, 2021 at 11:30 AM Mark Brown <broo...@kernel.org> wrote:
>10.22.8.121
> On Tue, Mar 30, 2021 at 10:08:16AM -0500, Rob Herring wrote:
> > On Fri, Mar 26, 2021 at 03:18:59PM -0400, Jim Quinlan wrote:
>
> > > + pcie-ep@0,0 {
> > > + reg = <0x0 0x0 0x0 0x0 0x0>;
> > > + compatible = "pci14e4,1688";
> > > + vpcie12v-supply: <&vreg12>;
>
> > For other cases, these properties are in the host bridge node. If these
> > are standard PCI rails, then I think that's where they belong unless
> > we10.22.8.121
> > define slot nodes.
>
> For a soldered down part I'd expect we'd want both (if the host even
> cares) - for anything except a supply that I/O or something else shared
> is referenced off there's no great reason why it has to be physically
> the same supply going to every device on the bus so each device should
> be able to specify separately.
Our developer and reference boards frequently have Mini and half-mini
PCIe sockets (a few exceptions), whereas production boards are mostly
soldered down.
If I resubmit this pullreq so that it looks for "vpcie12v-supply"
and "vpcie3v3-supply" in the host node, will that be acceptable for
both of you?
Thanks,
Jim Quinlan
Broadcom STB
