On Wed, 7 Oct 2026, Mert Erdoğan via B4 Relay wrote:

> From: Mert Erdoğan <mrterdo75@gmail.com>
> 
> The Victus 16-d1xxx ships with at least two DMI board names. 8A25 is
> already listed in victus_thermal_profile_boards, but 8A26 is not, so
> on those machines the driver falls back to the generic thermal policy.
> Selecting a platform profile then never issues the Victus performance
> mode WMI call (HPWMI_SET_PERFORMANCE_MODE via HPWMI_GM), and the
> firmware keeps the GPU's +15 W configurable TGP offset disabled.
> 
> Add 8A26 to the list so these machines use the same Victus thermal
> profile path as 8A25.
> 
> Before this change I enabled the cTGP offset on this machine with a
> userspace workaround that issues the same WMI call through acpi_call
> whenever power-profiles-daemon switches profile, and I still use that
> workaround on my daily kernel until this is merged.

First of all, thanks for the patch.

For changelog text, could you please try to write only what is relevant 
enough to be recorded into git history (it needs to be explanation how 
the patch came to be but focus on what the change does and why). If you 
want to add extra details like this, you can put them underneath the first 
--- line and our tools will cut that part out while the patch is being 
applied.

> Tested on a Victus by HP Laptop 16-d1xxx (board 8A26, BIOS F.22,
> RTX 3050 Laptop GPU) running 7.2.9, with the patched hp-wmi module
> built out of tree, the acpi_call workaround stopped and AC connected.
> The GPU power limit reported by nvidia-smi follows the selected
> profile, with the same values the acpi_call workaround produces:
> 
>   performance  75 W
>   balanced     60 W
>   quiet        30 W
> 
> Unplugging and replugging AC with the performance profile selected
> gives 20 W on battery and 75 W again on AC, again matching the
> workaround. The power source event handler reads the profile back
> from the EC without errors on both transitions.
> 
> With the unpatched module and the workaround stopped, switching
> between performance and balanced leaves the limit at 60 W, so the
> change in behaviour comes from this patch.
> 
> victus_thermal_profile_boards is identical in for-next and in 7.2.9,
> so the tested change is the same one-line change as below.
>
> This patch was prepared with an AI coding agent (Claude Code, using
> Claude Opus 5.5). It traced the driver's Victus code path and found
> it sends the same WMI call as my acpi_call workaround, wrote the
> change and this changelog, built the module and wrote the test
> script.

This and the previous paragraph too seems just unnecessary detai to 
recorded into git history and a bit repetitive as well.

Please also check the other change similarly and send these both once 
you're ready in a v2 series.

--
 i.

> I ran the test on the hardware and checked the results.
> 
> Assisted-by: claude-opus-5-5
> Signed-off-by: Mert Erdoğan <mrterdo75@gmail.com>
> ---
>  drivers/platform/x86/hp/hp-wmi.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/platform/x86/hp/hp-wmi.c b/drivers/platform/x86/hp/hp-wmi.c
> index 387f60e88..15f1e9f63 100644
> --- a/drivers/platform/x86/hp/hp-wmi.c
> +++ b/drivers/platform/x86/hp/hp-wmi.c
> @@ -234,6 +234,7 @@ static const char * const omen_timed_thermal_profile_boards[] = {
>  static const char * const victus_thermal_profile_boards[] = {
>  	"88F8",
>  	"8A25",
> +	"8A26",
>  };
>  
>  /* DMI board-specific feature data for Omen and Victus laptops. */
> 
> ---
> base-commit: 98f039265e580291d5e76a11b2072541fbb1c393
> change-id: 20261007-hp-wmi-victus-8a26-119e000cbe86
> 
> Best regards,
> --  
> Mert Erdoğan <mrterdo75@gmail.com>
> 
> 

-- 
 i.