[HELP] ipmi-kcs didn't work

[Vijay Khemka]

Hi Jay,
I found issue, As lpc-ctrl driver probe is failing because I didn't define flash and memory-region in DT for lpc-ctrl which probe look for. By defining these lpc-ctrl driver comes up and enable lpc clock. It is working for me now with this DT change.

Regards
-Vijay

On 12/3/18, 7:54 AM, "Jae Hyun Yoo" <jae.hyun.yoo at linux.intel.com> wrote:

    On 12/1/2018 8:29 AM, Samuel Jiang wrote:
    > Apologize for sending out no content mail first.
    > 
    > Jae,
    > The aspeed_gates in clk-aspeed.c perhaps as todo suggest asking Aspeed 
    > the actual parent data for check initializing?
    > 
    
    Yes, that makes sense.
    
    Hi Gary and Ryan,
    
    Can you please tell us what is the actual parent clock source of LPC IP?
    I mean the operational clock of LPC IP hardware block, not the interface
    clock.
    
    Thanks,
    Jae
    
    > Vijay,
    > I traced lpc-ctrl module, it seems direct update the same 
    > ASPEED_CLK_GATE_LCLK register map bit to enable. If parent data is 
    > disabled, it could enable in dts.
    > The device tree detail, I reference it from aspeed-g5.dtsi. Hope it 
    > could help you for work.
    > 
    > Thanks,
    > 
    > Samuel Jiang
    > On Dec 1, 2018, 7:04 AM +0800, Jae Hyun Yoo , wrote:
    >> On 11/30/2018 4:39 PM, Vijay Khemka wrote:
    >>>
    >>>
    >>> On 11/30/18, 11:29 AM, "Jae Hyun Yoo" <jae.hyun.yoo at linux.intel.com> 
    >>> wrote:
    >>>
    >>> On 11/30/2018 12:57 PM, Vijay Khemka wrote:
    >>>>
    >>>> On 11/30/18, 7:31 AM, "Jae Hyun Yoo" <jae.hyun.yoo at linux.intel.com> 
    >>>> wrote:
    >>>>
    >>>> On 11/29/2018 9:00 PM, Samuel Jiang wrote:
    >>>>> Hi Jae,
    >>>>>
    >>>>> We also encountered the same situation a month ago.
    >>>>> We try to enable the LPC clock on devicetree to resolve this 
    >>>>> problem not
    >>>>> direct modify clock driver.
    >>>>>
    >>>>> lpc_ctrl: lpc-ctrl at 0 {
    >>>>>   compatible = "aspeed,ast2500-lpc-ctrl”;
    >>>>>   reg = <0x0 0x80>;
    >>>>>   clocks = <&syscon ASPEED_CLK_GATE_LCLK>;
    >>>>> status = “okay”;
    >>>>> };
    >>>>>
    >>>>> However, I’m not sure which is better for LPC_CLCK setting.
    >>>>> Whether we should modify driver if original data is enabled or should
    >>>>> enable on the device tree, correct?
    >>>>>
    >>>>>
    >>>>
    >>>> Hi Samuel,
    >>>>
    >>>> Yes, you are right. I realized that kcs_bmc_aspeed.c doesn't have any
    >>>> clock control code so adding a clocks setting into kcs dtsi code doesn't
    >>>> make any help. Also, it was the reason why the change is needed on
    >>>> clk-aspeed.c. Enabling lpc_ctrl node you suggested seems a right way.
    >>>>
    >>>> Hi Vijay,
    >>>> Can you please try the way Samuel suggested after removing changes I
    >>>> gave you. Samuel's solution looks good to me.
    >>>>
    >>>> Thanks,
    >>>> Jae
    >>>>
    >>>> Jae,
    >>>> I can try. We can certainly enable lpc_ctrl node but in my 
    >>>> understanding that
    >>>> we still need this below patch to redefine ASPEED_CLK_GATE_LCLK value to
    >>>> add CRITICAL. I might be missing something here.
    >>>> - [ASPEED_CLK_GATE_LCLK] = { 8, 5, "lclk-gate", NULL, 0 }, /* LPC */
    >>>> + [ASPEED_CLK_GATE_LCLK] = { 8, 5, "lclk-gate", NULL, 
    >>>> CLK_IS_CRITICAL }, /* LPC */
    >>>>
    >>>
    >>> Vijay,
    >>>
    >>> I checked that lpc_ctrl module has clocks dt property handling code in
    >>> the module so I think the LCLK would be enabled without the clk-aspeed.c
    >>> fix. I can't test it now because I'm in a business trip so please test
    >>> it. I'll also check it when I back to office a week after.
    >>>
    >>> Thanks,
    >>> Jae
    >>>
    >>> Jae,
    >>> I removed above aspeed clock driver patch, and enabled lpc-ctrl node 
    >>> in dts also removed clock definition from kcs. And I don't see host 
    >>> is getting powered on. It works only with clock driver change.
    >>>
    >>> And I still didn't understand how enabling lpc-ctrl node will 
    >>> initialize ASPEED_CLK_GATE_LCLK to CRITICAL.
    >>>
    >>> -Vijay
    >>>
    >>
    >> Vijay,
    >>
    >> Thanks Vijay for sharing the test result. I'll also check it on my
    >> hardware when I get back to office. Will share my test result too.
    >>
    >> Thanks again,
    >> Jae