Re: [PATCH v3] platform/x86: hp-wmi: add support for OMEN 16-am2xxx boards

[Shuwen Bao]

On Tue, 27 Oct 2026 Ilpo Järvinen wrote:
> On Sat, 24 Oct 2026, Shuwen Bao wrote:
> 
> Hi,
> 
> It certainly sounds like a similar problem. I'm not sure if the code
> related to omen_timed_thermal_profile_boards is hooked though with
> this board because it requires is_omen_thermal_profile() returning
> true.
> 
> Some other drivers do profile keepalive by writing the same profile
> periodically if nothing has changed the profile for a while, that
> would seem easier/more generic solution than finding timer offset the
> current solution uses.
> 
> If you need to change something it would be nice to generalize this
> so that the keepalive is not bound to a particular submodel category.
> The driver is slowly moving towards that for all models so that we'd
> have only one central dmi matching table where the struct would hold
> all necessary info/quirk details.
> 
> > @Ilpo, since commit 2cd10a35f3f0 ("platform/x86: hp-wmi: add support
> > for OMEN 16-am2xxx boards") is already in review-ilpo-next, would
> > you prefer to revert it for now until I figure out a proper
> > solution, or should we keep it as-is while I investigate?
> 
> I think I'll keep the change as it still works to an extent so it
> seems better than nothing.

Hi Ilpo, Krishna,

Thank you for the guidance. I'll look into generalizing the keepalive 
mechanism as suggested.
In the meantime, I have observed a new symptom regarding this behavior: 
if no heavy load is applied to the system, the ~5000MHz boost clock 
(which only occurs under performance mode) can still be observed even 
after several hours.
It is currently still unclear whether this automatic reversion under 
load is being enforced directly by the EC firmware or handled somewhere 
within hp_wmi.c. I will continue investigating to determine the root 
cause.

Best regards,
Shwen Bao