On Tue, Jul 25, 2023 at 07:31:55AM +0200, Alexander Stein wrote: > Am Montag, 24. Juli 2023, 20:26:15 CEST schrieb Conor Dooley: > > On Mon, Jul 24, 2023 at 02:21:00PM +0200, Alexander Stein wrote: > > > i.MX8 and i.MX8X both use two clocks for accessing the periphery. > > > Add clocks and clock-names properties accordingly. > > > > > > Signed-off-by: Alexander Stein <alexander.stein@xxxxxxxxxxxxxxx> > > > --- > > > Changes in v2: > > > * None > > > > > > .../devicetree/bindings/media/nxp,imx8-jpeg.yaml | 8 ++++++++ > > > 1 file changed, 8 insertions(+) > > > > > > diff --git a/Documentation/devicetree/bindings/media/nxp,imx8-jpeg.yaml > > > b/Documentation/devicetree/bindings/media/nxp,imx8-jpeg.yaml index > > > 3d9d1db37040..2533e16720f2 100644 > > > --- a/Documentation/devicetree/bindings/media/nxp,imx8-jpeg.yaml > > > +++ b/Documentation/devicetree/bindings/media/nxp,imx8-jpeg.yaml > > > > > > @@ -46,6 +46,14 @@ properties: > > > minItems: 2 # Wrapper and 1 slot > > > maxItems: 5 # Wrapper and 4 slots > > > > > > + clocks: > > > + maxItems: 2 > > > + > > > + clock-names: > > > + items: > > > + - const: per > > > + - const: ipg > > > > What do "per" and "ipg" mean? I assume "per" is peripheral? > > Actually I don't know what "ipg" stands for. It's a quite common name on i.MX > platforms though. I opted for the names currently used in the DT. The driver > doesn't care for the names currently. Those names date back about 25 years to Motorola Mcore GSM SoCs. IPG came from IPG bus which IIRC stood for IP gasket. Essentially the bus was something like Arm APB being slave only. The IPG clock is essentially the bus and register access clock. 'per' is the functional clock in cases that need a defined clock rate such as UART baud clock. There is also a shared (between CPU and DSP) bus called SPBA from the same time which still lives on even though it isn't shared in i.MX chips. > But cross-checking the reference manual these clocks seems to be called "jpeg" > and "ips", individually for both jpeg encoder and decoder. Given this block is probably licensed IP, seems like it would use something different and be directly connected to AHB or AXI. > Mirela (added to recipients): As the original author of the DT nodes, could > you provide additional information regarding the clock names? > > Best regards, > Alexander
