Re: [PATCH v5] platform/x86: dell-laptop: Implement platform_profile

["Limonciello, Mario" <mario.limonciello@amd.com>]

On 5/12/2024 12:53 PM, Armin Wolf wrote:
> Am 11.05.24 um 17:56 schrieb Shen, Yijun:
> 
>> Internal Use - Confidential
>>> -----Original Message-----
>>> From: Limonciello, Mario <mario.limonciello@amd.com>
>>> Sent: Saturday, May 11, 2024 11:13 PM
>>> To: Shen, Yijun <Yijun_Shen@Dell.com>; Lyndon Sanche
>>> <lsanche@lyndeno.ca>
>>> Cc: pali@kernel.org; W_Armin@gmx.de;
>>> srinivas.pandruvada@linux.intel.com; ilpo.jarvinen@linux.intel.com;
>>> lkp@intel.com; Hans de Goede <hdegoede@redhat.com>; Matthew Garrett
>>> <mjg59@srcf.ucam.org>; Jonathan Corbet <corbet@lwn.net>; Heiner Kallweit
>>> <hkallweit1@gmail.com>; Vegard Nossum <vegard.nossum@oracle.com>;
>>> platform-driver-x86@vger.kernel.org; linux-kernel@vger.kernel.org; 
>>> Dell Client
>>> Kernel <Dell.Client.Kernel@dell.com>
>>> Subject: Re: [PATCH v5] platform/x86: dell-laptop: Implement 
>>> platform_profile
>>>
>>>
>>> [EXTERNAL EMAIL]
>>>
>>>
>>>
>>> On 5/11/2024 10:05 AM, Shen, Yijun wrote:
>>>> Internal Use - Confidential
>>>>> -----Original Message-----
>>>>> From: Mario Limonciello <mario.limonciello@amd.com>
>>>>> Sent: Wednesday, May 8, 2024 11:53 PM
>>>>> To: Shen, Yijun <Yijun_Shen@Dell.com>; Lyndon Sanche
>>>>> <lsanche@lyndeno.ca>
>>>>> Cc: pali@kernel.org; W_Armin@gmx.de;
>>>>> srinivas.pandruvada@linux.intel.com; ilpo.jarvinen@linux.intel.com;
>>>>> lkp@intel.com; Hans de Goede <hdegoede@redhat.com>; Matthew
>>> Garrett
>>>>> <mjg59@srcf.ucam.org>; Jonathan Corbet <corbet@lwn.net>; Heiner
>>>>> Kallweit <hkallweit1@gmail.com>; Vegard Nossum
>>>>> <vegard.nossum@oracle.com>; platform-driver-x86@vger.kernel.org;
>>>>> linux-kernel@vger.kernel.org; Dell Client Kernel
>>>>> <Dell.Client.Kernel@dell.com>
>>>>> Subject: Re: RE: [PATCH v5] platform/x86: dell-laptop: Implement
>>>>> platform_profile
>>>>>
>>>>>
>>>>> [EXTERNAL EMAIL]
>>>>>
>>>>> On 5/8/2024 09:24, Shen, Yijun wrote:
>>>>>> Hi Lyndon,
>>>>>>
>>>>>>     Thanks for working on this patch.
>>>>>>
>>>>>>     Dell side has an initial testing with this patch on some laptops,
>>>>>> it looks
>>>>> good. While changing the platform profile:
>>>>>> 1. The corresponding USTT option in BIOS will be changed.
>>>>>> 2. thermald will not be impacted. The related PSVT and ITMT will be
>>> loaded.
>>>>>>     Some Dell DTs does not have the USTT, Dell'll have a check if
>>>>>> nothing is
>>>>> broken.
>>>>>
>>>>> Hi Alex!
>>>>>
>>>>> Have you had a check both on both your AMD laptops and workstations
>>>>> too, or just the Intel ones?  I think it would be good to make sure
>>>>> it's getting the correct experience in both cases.
>>>>>
>>>> Hi Mario,
>>>>
>>>>    I've a check for this, for both laptop and workstation, the 
>>>> dell_laptop
>>> module will not be loaded. So, AMD platform will not be impacted by this
>>> patch series.
>>>> Follow is one example output with workstation.
>>>>    #lsmod | grep dell
>>>>      dell_wmi               28672  0
>>>>      dell_smbios            32768  1 dell_wmi
>>>>      dcdbas                 20480  1 dell_smbios
>>>>      dell_wmi_descriptor    20480  2 dell_wmi,dell_smbios
>>>>      sparse_keymap          12288  1 dell_wmi
>>>>      ledtrig_audio          12288  3 
>>>> snd_ctl_led,snd_hda_codec_generic,dell_wmi
>>>>      video                  73728  2 dell_wmi,nvidia_modeset
>>>>      wmi                    40960  5
>>> video,dell_wmi,wmi_bmof,dell_smbios,dell_wmi_descriptor
>>> Ah; right that makes sense.  In that case, is dell-laptop even the 
>>> right place for
>>> this patch series?  I would think the same policies for the platform 
>>> profile
>>> should be able to apply to desktop/workstation.
>>>
>>> The v6 of this series would block smbios-thermal-ctl from working on a
>>> workstation too.
>>>
>>>>>>      Additional, with this patch, follow behavior is found:
>>>>>>     1. For example, the platform profile is quiet.
>>>>>>     2. Reboot the system and change the USTT to performance.
>>>>>>     3. Boot to desktop, the platform profile is "quiet", the USTT
>>>>>> will be
>>>>> changed back to "quiet".
>>>>>>     This looks like not a proper user experience. The platform
>>>>>> profile should
>>>>> honor the BIOS setting, aka, the platform profile should be switched
>>>>> to "performance".
>>>>> I agree, this sounds like the initial profile needs to be read from
>>>>> the BIOS settings too.
>>>>>
>>>>> Furthermore I wanted to ask is there also a WMI setting that
>>>>> corresponds to this that dell-wmi-sysman offers?
>>>>    Yes, Mario, you're right. This thermal setting could also be 
>>>> toggled by dell-
>>> wmi-sysman.
>>>> But, for the Dell consumer AMD laptops, like Alienware, the BIOS is 
>>>> another
>>> variant which is different with the workstation one.
>>>> With this variant BIOS, there is no USTT and also no dell_wmi/dell-wmi-
>>> sysman.
>>>>> I'm wondering if both should be probed in case the SMBIOS one goes
>>> away one day.
>>>>    Yep, I think this is a good suggestion.
>>>>
>>> Great! Although something I wonder is if the policy when changed with 
>>> dell-
>>> wmi-sysman is immediate or requires a reboot.  A lot of the settings 
>>> in there
>>> aren't effective until after a reboot.
>>>
>>> If this is one of them then it might not be a good idea to make it 
>>> work for
>>> both.
>> Hi Mario,
>>
>>   Just have a check, the check steps are:
>> 1. stop the thermald
>> 2. run the stress test
>> 3. Toggle the thermal setting between UltraPerformance and Quiet via 
>> dell-wmi-sysman
>> 4. Check the CPU FAN speed
>>   The system reboot is not needed, the CPU fan speed changes immediately.
>>   A screen recorder here: 
>> https://dell.box.com/s/p2bhd2b6cv8z5buk9eao3bosgrrp1lf9
>>
>> Thanks
>>
> Hi,
> 
> i believe it should be the responsibility of the manufacturer (in this 
> case Dell) that
> the thermal state remains consistent across both interfaces.
> 
> I think that the official Windows utility only checks the thermal state 
> reported by
> the USTT interface, so we should match this behavior.
> 
> Thanks,
> Armin Wolf
> 

Why?  Windows also does ACPI-WMI differently than Linux.  It's not as 
easy to check both from a Windows utility due to that.