Re: [PATCH] cpufreq/amd-pstate: Supply nominal/lowest freq for TRX40-based motherboards

[Mario Limonciello]

+ Kyle

Kyle,

By chance is this the same system that you were talking to me about offline?



On 10/7/26 09:41, Giovanni Gherdovich wrote:
> Some motherboard based on the TRX40 chipset has been reported to ship
> firmware lacking lowest and nominal frequency values in the
> _CPC ACPI package. That is, it supports CPPC V2 but not CPPC V3.
> Firmware updates aren't available.

Can you please add more information about the vendor/model of the MB, etc?
> amd-pstate requires explicit knowledge of nominal frequency, so it
> can't load on this hardware. The driver already has a mechanism (the
> so-called "quirks") to accommodate for missing nominal freq in ACPI
> tables, so here we use it to match against CPU family, model, core
> count, and BIOS version.

Yeah; it's intended for this specific case of really old hardware that 
the BIOS isn't going to fix it.

I don't understand why core count matters though.

> Signed-off-by: Giovanni Gherdovich <ggherdovich@xxxxxxx>
> ---
>   drivers/cpufreq/amd-pstate.c | 31 +++++++++++++++++++++++++++++++
>   1 file changed, 31 insertions(+)
>
> diff --git a/drivers/cpufreq/amd-pstate.c b/drivers/cpufreq/amd-pstate.c
> index 8bfd46d60843..532ff16dd076 100644
> --- a/drivers/cpufreq/amd-pstate.c
> +++ b/drivers/cpufreq/amd-pstate.c
> @@ -145,6 +145,11 @@ static struct quirk_entry quirk_amd_7k62 = {
>   	.lowest_freq = 550,
>   };
>   
> +static struct quirk_entry quirk_amd_ryzen_threadripper_3000_24c = {
> +	.nominal_freq = 3800,
> +	.lowest_freq = 550,
> +};
> +
>   static inline u8 freq_to_perf(union perf_cached perf, u32 nominal_freq, unsigned int freq_val)
>   {
>   	u32 perf_val = DIV_ROUND_UP_ULL((u64)freq_val * perf.nominal_perf, nominal_freq);
> @@ -174,6 +179,22 @@ static int __init dmi_matched_7k62_bios_bug(const struct dmi_system_id *dmi)
>   	return 0;
>   }
>   
> +static int __init dmi_matched_trx40_bios_bug(const struct dmi_system_id *dmi)
> +{
> +	/**
> +	 * Match the Ryzen Threadripper 3000 series, 24-core SKU, sTRX4 socket / TRX40 chipset.
> +	 */
> +	if (boot_cpu_data.x86 == 0x17 &&
> +	    boot_cpu_data.x86_model >= 0x30 && boot_cpu_data.x86_model <= 0x3F &&
> +	    topology_num_cores_per_package() == 24) {

Does the number of cores actually matter?  Do you mean to say if you 
swap the CPU to another part CPPC works?

> +		quirks = dmi->driver_data;
> +		pr_info("Overriding nominal and lowest frequencies for %s\n", dmi->ident);

The BIOS bug specifically is lack of values, not invalid values, right? 
Just want to make sure I'm following this right.

> +		return 1;
> +	}
> +
> +	return 0;
> +}
> +
>   static const struct dmi_system_id amd_pstate_quirks_table[] __initconst = {
>   	{
>   		.callback = dmi_matched_7k62_bios_bug,
> @@ -184,6 +205,16 @@ static const struct dmi_system_id amd_pstate_quirks_table[] __initconst = {
>   		},
>   		.driver_data = &quirk_amd_7k62,
>   	},
> +	{
> +		.callback = dmi_matched_trx40_bios_bug,
> +		.ident = "AMD Ryzen Threadripper 3000",
> +		.matches = {
> +			DMI_MATCH(DMI_BIOS_VENDOR,  "American Megatrends International"),
> +			DMI_MATCH(DMI_BIOS_VERSION, "2.80"),
> +			DMI_MATCH(DMI_BIOS_DATE,    "05/17/2022"),
> +		},
> +		.driver_data = &quirk_amd_ryzen_threadripper_3000_24c,
> +	},
>   	{}
>   };
>   MODULE_DEVICE_TABLE(dmi, amd_pstate_quirks_table);