On Thu, Sep 12, 2019 at 10:46:58PM +0200, Jernej Škrabec wrote:
> Dne četrtek, 12. september 2019 ob 22:34:27 CEST je Maxime Ripard napisal(a):
> > On Thu, Sep 12, 2019 at 10:28:37PM +0200, Jernej Škrabec wrote:
> > > Dne četrtek, 12. september 2019 ob 22:20:57 CEST je Maxime Ripard
> napisal(a):
> > > > Hi,
> > > >
> > > > On Thu, Sep 12, 2019 at 07:51:29PM +0200, Jernej Skrabec wrote:
> > > > > Both, H3 and H5, contain MBUS, which is the bus used by DMA devices to
> > > > > access system memory.
> > > > >
> > > > > MBUS controller is responsible for arbitration between channels based
> > > > > on set priority and can do some other things as well, like report
> > > > > bandwidth used. It also maps RAM region to different address than CPU.
> > > > >
> > > > > Signed-off-by: Jernej Skrabec <jernej.skrabec@xxxxxxxx>
> > > > > ---
> > > > >
> > > > > arch/arm/boot/dts/sunxi-h3-h5.dtsi | 9 +++++++++
> > > > > 1 file changed, 9 insertions(+)
> > > > >
> > > > > diff --git a/arch/arm/boot/dts/sunxi-h3-h5.dtsi
> > > > > b/arch/arm/boot/dts/sunxi-h3-h5.dtsi index eba190b3f9de..ef1d03812636
> > > > > 100644
> > > > > --- a/arch/arm/boot/dts/sunxi-h3-h5.dtsi
> > > > > +++ b/arch/arm/boot/dts/sunxi-h3-h5.dtsi
> > > > > @@ -109,6 +109,7 @@
> > > > >
> > > > > compatible = "simple-bus";
> > > > > #address-cells = <1>;
> > > > > #size-cells = <1>;
> > > > >
> > > > > + dma-ranges;
> > > > >
> > > > > ranges;
> > > > >
> > > > > display_clocks: clock@1000000 {
> > > > >
> > > > > @@ -538,6 +539,14 @@
> > > > >
> > > > > };
> > > > >
> > > > > };
> > > > >
> > > > > + mbus: dram-controller@1c62000 {
> > > > > + compatible = "allwinner,sun8i-h3-mbus";
> > > > > + reg = <0x01c62000 0x1000>;
> > > > > + clocks = <&ccu 113>;
> > > > > + dma-ranges = <0x00000000 0x40000000
> > >
> > > 0xc0000000>;
> > >
> > > > > + #interconnect-cells = <1>;
> > > > > + };
> > > > > +
> > > >
> > > > If that's easy enough to access, can you also add the references in
> > > > the devices that are already there? (CSI and DE comes to my mind, but
> > > > there might be others).
> > >
> > > Strangely, DE2 doesn't use this offset. That was tested on OrangePi
> > > Plus2E,
> > > which has 2 GiB of RAM and subtracting this offset causes corrupted image.
> >
> > Ok, weird. But if it was tested then fine by me :)
> >
> > > But I can add this properties to CSI too. However, wouldn't that need CSI
> > > DT binding expansion with those properties? othetwise DT check will fail.
> > Oh right, we definitely need to update the binding indeed. The code
> > should be able to cope with both cases already.
>
> I guess it's better to handle that with another patch series then? Changing
> CSI bindings doesn't fit here.
Yeah, you can do it in a separate series if you prefer
Maxime
