Hi,
On 9/6/2023 8:21 PM, Jason Andryuk wrote:
> From: Roger Pau Monne <roger.pau@xxxxxxxxxx>
>
> The Processor _PDC buffer bits notify ACPI of the OS capabilities, and
> so ACPI can adjust the return of other Processor methods taking the OS
> capabilities into account.
_PDC method is deprecated for this purpose, since 2018, and is dropped from
spec since 6.5
We made the switch in linux since 6.6:
95272641338a ("ACPI: processor: Use _OSC to convey OSPM processor support information")
>
> When Linux is running as a Xen dom0, it's the hypervisor the entity
> in charge of processor power management, and hence Xen needs to make
> sure the capabilities reported in the _PDC buffer match the
> capabilities of the driver in Xen.
So I guess you would need to sanitize buffer passed to _OSC method instead ?
>
> Introduce a small helper to sanitize the buffer when running as Xen
> dom0.
>
> When Xen supports HWP, this serves as the equivalent of commit
> a21211672c9a ("ACPI / processor: Request native thermal interrupt
> handling via _OSC") to avoid SMM crashes. Xen will set bit 12 in the
> _PDC bits and the _PDC call will apply it.
>
> [ jandryuk: Mention Xen HWP's need. Move local variables ]
> Signed-off-by: Roger Pau Monné <roger.pau@xxxxxxxxxx>
> Cc: stable@xxxxxxxxxxxxxxx
> Signed-off-by: Jason Andryuk <jandryuk@xxxxxxxxx>
> ---
> v2:
> Move local variables in acpi_processor_eval_pdc() to reuse in both conditions.
> ---
> arch/x86/include/asm/xen/hypervisor.h | 6 ++++++
> arch/x86/xen/enlighten.c | 19 +++++++++++++++++++
> drivers/acpi/processor_pdc.c | 22 ++++++++++++++++------
> 3 files changed, 41 insertions(+), 6 deletions(-)
>
> diff --git a/arch/x86/include/asm/xen/hypervisor.h b/arch/x86/include/asm/xen/hypervisor.h
> index 5fc35f889cd1..0f88a7e450d3 100644
> --- a/arch/x86/include/asm/xen/hypervisor.h
> +++ b/arch/x86/include/asm/xen/hypervisor.h
> @@ -63,4 +63,10 @@ void __init xen_pvh_init(struct boot_params *boot_params);
> void __init mem_map_via_hcall(struct boot_params *boot_params_p);
> #endif
>
> +#ifdef CONFIG_XEN_DOM0
> +void xen_sanitize_pdc(uint32_t *buf);
> +#else
> +static inline void xen_sanitize_pdc(uint32_t *buf) { BUG(); }
> +#endif
> +
> #endif /* _ASM_X86_XEN_HYPERVISOR_H */
> diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
> index b8db2148c07d..9f7fc11330a3 100644
> --- a/arch/x86/xen/enlighten.c
> +++ b/arch/x86/xen/enlighten.c
> @@ -346,3 +346,22 @@ void xen_arch_unregister_cpu(int num)
> }
> EXPORT_SYMBOL(xen_arch_unregister_cpu);
> #endif
> +
> +#ifdef CONFIG_XEN_DOM0
> +void xen_sanitize_pdc(uint32_t *buf)
> +{
> + struct xen_platform_op op = {
> + .cmd = XENPF_set_processor_pminfo,
> + .interface_version = XENPF_INTERFACE_VERSION,
> + .u.set_pminfo.id = -1,
> + .u.set_pminfo.type = XEN_PM_PDC,
> + };
> + int ret;
> +
> + set_xen_guest_handle(op.u.set_pminfo.pdc, buf);
> + ret = HYPERVISOR_platform_op(&op);
> + if (ret)
> + pr_info("sanitize of _PDC buffer bits from Xen failed: %d\n",
> + ret);
> +}
> +#endif
> diff --git a/drivers/acpi/processor_pdc.c b/drivers/acpi/processor_pdc.c
> index 18fb04523f93..9393dd4a3158 100644
> --- a/drivers/acpi/processor_pdc.c
> +++ b/drivers/acpi/processor_pdc.c
> @@ -122,6 +122,11 @@ static acpi_status
> acpi_processor_eval_pdc(acpi_handle handle, struct acpi_object_list *pdc_in)
> {
> acpi_status status = AE_OK;
> + union acpi_object *obj;
> + u32 *buffer = NULL;
> +
> + obj = pdc_in->pointer;
> + buffer = (u32 *)(obj->buffer.pointer);
>
> if (boot_option_idle_override == IDLE_NOMWAIT) {
> /*
> @@ -129,14 +134,19 @@ acpi_processor_eval_pdc(acpi_handle handle, struct acpi_object_list *pdc_in)
> * mode will be disabled in the parameter of _PDC object.
> * Of course C1_FFH access mode will also be disabled.
> */
> - union acpi_object *obj;
> - u32 *buffer = NULL;
> -
> - obj = pdc_in->pointer;
> - buffer = (u32 *)(obj->buffer.pointer);
> buffer[2] &= ~(ACPI_PDC_C_C2C3_FFH | ACPI_PDC_C_C1_FFH);
> -
> }
> +
> + if (xen_initial_domain()) {
> + /*
> + * When Linux is running as Xen dom0, the hypervisor is the
> + * entity in charge of the processor power management, and so
> + * Xen needs to check the OS capabilities reported in the _PDC
> + * buffer matches what the hypervisor driver supports.
> + */
> + xen_sanitize_pdc(buffer);
> + }
> +
> status = acpi_evaluate_object(handle, "_PDC", pdc_in, NULL);
>
> if (ACPI_FAILURE(status))
