[PATCH v2] clk: aspeed: Mark lclk (LPC) and espiclk (eSPI) as critical
Jae Hyun Yoo
jae.hyun.yoo at linux.intel.com
Thu Dec 6 02:04:30 AEDT 2018
On 12/4/2018 2:54 PM, Stephen Boyd wrote:
> Quoting Joel Stanley (2018-12-04 12:51:43)
>> Hi Jae,
>>
>> On Wed, 5 Dec 2018 at 05:20, Jae Hyun Yoo <jae.hyun.yoo at linux.intel.com> wrote:
>>>
>>> These interfaces are used by host to talk to BMC, and the clock
>>> source is from the host, usually from PCH. So this commit marks
>>> the lclk as critical to make it able to be enabled. Also, it marks
>>> espiclk too because eSPI is sharing the same interface with LPC.
>>
>> While this is true on the platform you have in mind, on other
>> platforms this is not the case. They do not use eSPI, and LCLK is
>> enabled by the driver that is used to configure LPC
>> (drivers/misc/aspeed-lpc-ctrl.c):
>>
>> lpc_ctrl: lpc-ctrl at 0 {
>> compatible = "aspeed,ast2400-lpc-ctrl";
>> reg = <0x0 0x80>;
>> clocks = <&syscon ASPEED_CLK_GATE_LCLK>;
>> }
>>
>> I wonder if we need a device tree binding to describe which clocks are critical.
>
> A binding to describe critical clks has been rejected in the past. I
> don't think we need to have it here either? More information on why
> things are being marked critical will be helpful to see if we need to
> re-open that discussion again.
>
>>
>> Stephen, please don't merge this patch yet.
>>
>
> Sure.
>
Hi Stephen and Joel,
Thanks for blocking this patch. We checked that LCLK gate can be opened
by enabling lpc-ctrl node without using this change. Please drop this
patch.
Thanks,
Jae
More information about the openbmc
mailing list