Re: [PATCH v5 2/5] Documentation: DT: Add bindings for Spreadtrum SoC Platform
From: Orson Zhai <hidden>
Date: 2015-01-17 08:10:38
Also in:
linux-arm-kernel, linux-devicetree, linux-serial, lkml
On Fri, Jan 16, 2015 at 10:11 PM, Mark Rutland [off-list ref] wrote:
On Fri, Jan 16, 2015 at 12:53:16PM +0000, Lyra Zhang wrote:quoted
Hi, Markquoted
quoted
+ +Required properties: +- compatible: must be "sprd,sc9836-uart" +- reg: offset and length of the register set for the device +- interrupts: exactly one interrupt specifier +- clocks: phandles to input clocks.The order and relevance of each should be specified. If you have multiple clocks I would strongly recommend you use clock-names to distinguish them.Thank you for the recommendation. but, since we haven't made the clock driver ready, for this initial commit, we just let 4 UARTs share a single fixed 26 MHz clock source. we'll do like you've recommended when we will submit the clock driver in the future.I'm on about the clock input lines on the UART instance, not the providers they come from. Is there only a single clock input line on each UART? Perhaps multiple input lines which are currently fed by the same clock?
________
| 26MHz |-------------------------------------------------
------------- | |
| |
_______ ________
| UART1 | | UART2 | .........
-------------- -------------
the hardware is something like the diagram.
4 Uart modules are all connected to a fixed 26Mhz crystal by power-on default.
There should be a clock-mux between uart and 26Mhz which
could select other clock source such as some pll output.
But as initial commit , we are not ready to describe other inputs by
these muxes.
So we treat the UART as a simple model with only one fixed-clock input.
And we plan to add the other inputs path back in a not very far future.
Is it appropriate to do like this?
Orson
Thanks, Mark.