Re: [PATCH v2 2/2] hwmon: add initial support for AMD PROM21 xHCI temperature sensor

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

On 5/7/26 04:08, Michal Pecio wrote:
> On Wed, 6 May 2026 16:36:49 -0500, Mario Limonciello wrote:
>>>> The temperature register did not return a valid value while the
>>>> xHCI PCI function was suspended in testing. Keep the existing
>>>> behavior by default and allow temperature reads to wake the xHCI
>>>> PCI device. Add an allow_pm_switch module parameter so users can
>>>> disable that behavior; when disabled, reads do not wake the
>>>> device and return -EAGAIN if it is suspended.
>>>
>>> Is such behavior useful?
>>>
>>> Maybe the driver could just disable runtime PM while it's loaded.
>>
>> I'd encourage what we do in amdgpu for dGPUs. The hwmon files will
>> return an error code (I forget which code) when the device is in
>> runtime PM when called.  Don't explicitly wake it otherwise.
>>
>> This prevents someone installing a sensor monitoring application and
>> that application "being the only thing" keeping the dGPU awake.  If
>> it's awake already for other reasons (like being used) then return
>> valid data to the applications
> 
> Well, that's not a dGPU but an xHCI controller embedded in the chipset,
> which chipset is more or less active all the time (includes bridges to
> PCIe ports, some SATA controllers and mabe other things I forgot).
> Is the saving from disabling xHCI significant for a desktop system?
> 
> Users may be interested in monitoring chipset temperature even while
> not actively using USB.
> 
> I don't know what are the conditions to put GPUs into runtime suspend,
> but a USB HC will be going in and out quite randomly, depending on
> connected devices and their workload. You may end up needing to answer
> people why their sensor only works when they turn on a webcam :)

Heh.  You have a good point here.  The reason for this policy in dGPUs 
came from a lot of people using them "headless" (for example plugging in 
a display to the iGPU on the board) and wondering why randomly their 
power shot up and fans spun up on the GPU.  It took a while to narrow 
down that it was specifically sensor software doing it.

For a sensor accessed through registers on XHCI this is a lot less of a 
big deal for all the reasons you point out.

> 
> Alternatively, would it be possible to bring a suspendend HC into D0,
> read the temperature register and then put it back into D3hot without
> bothering the USB layer to fully resume and suspend xHCI logic?
> 
This sounds better to me if it's feasible.