Re: [PATCH] remoteproc: xlnx: avoid RPU force power down

[Tanmay Shah <tanmay.shah@amd.com>]

On 4/22/25 2:10 PM, Mathieu Poirier wrote:
> On Tue, 22 Apr 2025 at 12:30, Tanmay Shah <tanmay.shah@amd.com> wrote:
>>
>>
>>
>> On 4/22/25 12:49 PM, Mathieu Poirier wrote:
>>> On Tue, 22 Apr 2025 at 10:10, Tanmay Shah <tanmay.shah@amd.com> wrote:
>>>>
>>>>
>>>>
>>>> On 4/22/25 10:59 AM, Mathieu Poirier wrote:
>>>>> Good morning,
>>>>>
>>>>> On Mon, Apr 14, 2025 at 11:46:01AM -0700, Tanmay Shah wrote:
>>>>>> Powering off RPU using force_pwrdwn call results in system failure
>>>>>> if there are multiple users of that RPU node. Better mechanism is to use
>>>>>> request_node and release_node EEMI calls. With use of these EEMI calls,
>>>>>> platform management controller will take-care of powering off RPU
>>>>>> when there is no user.
>>>>>>
>>>>>> Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
>>>>>> ---
>>>>>>     drivers/remoteproc/xlnx_r5_remoteproc.c | 29 ++++++++++++++++++++++++-
>>>>>>     1 file changed, 28 insertions(+), 1 deletion(-)
>>>>>>
>>>>>> diff --git a/drivers/remoteproc/xlnx_r5_remoteproc.c b/drivers/remoteproc/xlnx_r5_remoteproc.c
>>>>>> index 5aeedeaf3c41..3597359c0fc8 100644
>>>>>> --- a/drivers/remoteproc/xlnx_r5_remoteproc.c
>>>>>> +++ b/drivers/remoteproc/xlnx_r5_remoteproc.c
>>>>>> @@ -380,6 +380,18 @@ static int zynqmp_r5_rproc_start(struct rproc *rproc)
>>>>>>        dev_dbg(r5_core->dev, "RPU boot addr 0x%llx from %s.", rproc->bootaddr,
>>>>>>                bootmem == PM_RPU_BOOTMEM_HIVEC ? "OCM" : "TCM");
>>>>>>
>>>>>> +    /* Request node before starting RPU core if new version of API is supported */
>>>>>> +    if (zynqmp_pm_feature(PM_REQUEST_NODE) > 1) {
>>>>>> +            ret = zynqmp_pm_request_node(r5_core->pm_domain_id,
>>>>>> +                                         ZYNQMP_PM_CAPABILITY_ACCESS, 0,
>>>>>> +                                         ZYNQMP_PM_REQUEST_ACK_BLOCKING);
>>>>>> +            if (ret < 0) {
>>>>>> +                    dev_err(r5_core->dev, "failed to request 0x%x",
>>>>>> +                            r5_core->pm_domain_id);
>>>>>> +                    return ret;
>>>>>> +            }
>>>>>> +    }
>>>>>> +
>>>>>>        ret = zynqmp_pm_request_wake(r5_core->pm_domain_id, 1,
>>>>>>                                     bootmem, ZYNQMP_PM_REQUEST_ACK_NO);
>>>>>>        if (ret)
>>>>>> @@ -401,10 +413,25 @@ static int zynqmp_r5_rproc_stop(struct rproc *rproc)
>>>>>>        struct zynqmp_r5_core *r5_core = rproc->priv;
>>>>>>        int ret;
>>>>>>
>>>>>> +    /* Use release node API to stop core if new version of API is supported */
>>>>>> +    if (zynqmp_pm_feature(PM_RELEASE_NODE) > 1) {
>>>>>> +            ret = zynqmp_pm_release_node(r5_core->pm_domain_id);
>>>>>> +            if (ret)
>>>>>> +                    dev_err(r5_core->dev, "failed to stop remoteproc RPU %d\n", ret);
>>>>>> +            return ret;
>>>>>> +    }
>>>>>> +
>>>>>> +    if (zynqmp_pm_feature(PM_FORCE_POWERDOWN) < 1) {
>>>>>> +            dev_dbg(r5_core->dev, "EEMI interface %d not supported\n",
>>>>>> +                    PM_FORCE_POWERDOWN);
>>>>>> +            return -EOPNOTSUPP;
>>>>>> +    }
>>>>>
>>>>> Here I have to guess, because it is not documented, that it is the check to see
>>>>> if zynqmp_pm_force_pwrdwn() is available.  I'm not sure why it is needed because
>>>>> zynqmp_pm_force_pwrdwn() returns and error code.
>>>>>
>>>> Hello,
>>>>
>>>> Thanks for reviews. Yes you are correct. Actually instead, the check
>>>> should be for version 1 of PM_FORCE_POWER_DOWN. If version 1 is
>>>> supported, only then execute the call otherwise print the error.
>>>> Hence, the check should be something like:
>>>>
>>>> if (zynqmp_pm_feature(PM_FORCE_POWERDOWN) != 1) {
>>>>           error out.
>>>> }
>>>>
>>>
>>> The above still doesn't answer my question, i.e _why_ is a check
>>> needed when zynqmp_pm_force_pwrdwn() returns an error code?  To me, if
>>> something happens in zynqmp_pm_force_pwrdwn() then an error code is
>>> reported and the current implementation is able to deal with it.
>>>
>>
>> PM_FORCE_POWERDOWN will print redundant error messages from firmware if
>> called for feature that is not supported. By doing above version check,
>> we are avoiding those unnecessary error/warning messages. Other than
>> that, you are correct we don't need to do version check as
>> PM_FORCE_POWERDOWN will send respective error code and we will fail
>> here. But version check helps to differentiate between actual error log
>> from firmware when call is expected to work.
>>
> 
> That is the kind of information that would be useful as comments in
> the code.  Otherwise there is simply no way to tell...
> 

Yes that makes sense. I will update comment accordingly.

>>>> I will fix and add comment as well.
>>>>
>>>>> Thanks,
>>>>> Mathieu
>>>>>
>>>>>> +
>>>>>> +    /* maintain force pwr down for backward compatibility */
>>>>>>        ret = zynqmp_pm_force_pwrdwn(r5_core->pm_domain_id,
>>>>>>                                     ZYNQMP_PM_REQUEST_ACK_BLOCKING);
>>>>>>        if (ret)
>>>>>> -            dev_err(r5_core->dev, "failed to stop remoteproc RPU %d\n", ret);
>>>>>> +            dev_err(r5_core->dev, "core force power down failed\n");
>>>>>>
>>>>>>        return ret;
>>>>>>     }
>>>>>>
>>>>>> base-commit: 8532691d0a85ab2a826808207e904f7d62a9d804
>>>>>> --
>>>>>> 2.34.1
>>>>>>
>>>>
>>