Re: [PATCH v4] acpi/processor: sanitize _OSC/_PDC capabilities for Xen dom0

["Roger Pau Monné" <roger.pau@citrix.com>]

On Wed, Nov 01, 2023 at 09:41:52AM -0400, Jason Andryuk wrote:
> From: Roger Pau Monne <roger.pau@citrix.com>
> 
> The Processor capability bits notify ACPI of the OS capabilities, and
> so ACPI can adjust the return of other Processor methods taking the OS
> capabilities into account.
> 
> When Linux is running as a Xen dom0, the hypervisor is the entity
> in charge of processor power management, and hence Xen needs to make
> sure the capabilities reported by _OSC/_PDC match the capabilities of
> the driver in Xen.
> 
> 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
> ACPI_PROC_CAP_COLLAB_PROC_PERF (bit 12) in the capability bits and the
> _OSC/_PDC call will apply it.
> 
> [ jandryuk: Mention Xen HWP's need.  Support _OSC & _PDC ]
> Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
> Cc: stable@vger.kernel.org
> Signed-off-by: Jason Andryuk <jandryuk@gmail.com>
> ---
> v4:
> Use xen_santize_proc_cap_bits() name - Michal
> Rephrase comment - Michal
> 
> v3:
> Move xen_sanitize_pdc() call to arch_acpi_set_proc_cap_bits() to cover
> _OSC and _PDC.
> drivers/xen/pcpu.c is CONFIG_DOM0 && CONFIG_X86
> 
> v2:
> Move local variables in acpi_processor_eval_pdc() to reuse in both conditions.
> ---
>  arch/x86/include/asm/acpi.h           | 14 ++++++++++++++
>  arch/x86/include/asm/xen/hypervisor.h |  9 +++++++++
>  drivers/xen/pcpu.c                    | 21 +++++++++++++++++++++
>  3 files changed, 44 insertions(+)
> 
> diff --git a/arch/x86/include/asm/acpi.h b/arch/x86/include/asm/acpi.h
> index c8a7fc23f63c..f896eed4516c 100644
> --- a/arch/x86/include/asm/acpi.h
> +++ b/arch/x86/include/asm/acpi.h
> @@ -16,6 +16,9 @@
>  #include <asm/x86_init.h>
>  #include <asm/cpufeature.h>
>  #include <asm/irq_vectors.h>
> +#include <asm/xen/hypervisor.h>
> +
> +#include <xen/xen.h>
>  
>  #ifdef CONFIG_ACPI_APEI
>  # include <asm/pgtable_types.h>
> @@ -127,6 +130,17 @@ static inline void arch_acpi_set_proc_cap_bits(u32 *cap)
>  	if (!cpu_has(c, X86_FEATURE_MWAIT) ||
>  	    boot_option_idle_override == IDLE_NOMWAIT)
>  		*cap &= ~(ACPI_PROC_CAP_C_C1_FFH | ACPI_PROC_CAP_C_C2C3_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
> +		 * processor capabilities buffer matches what the hypervisor
> +		 * driver supports.
> +		 */
> +		xen_sanitize_proc_cap_bits(cap);
> +	}
>  }
>  
>  static inline bool acpi_has_cpu_in_madt(void)
> diff --git a/arch/x86/include/asm/xen/hypervisor.h b/arch/x86/include/asm/xen/hypervisor.h
> index 7048dfacc04b..a9088250770f 100644
> --- a/arch/x86/include/asm/xen/hypervisor.h
> +++ b/arch/x86/include/asm/xen/hypervisor.h
> @@ -100,4 +100,13 @@ static inline void leave_lazy(enum xen_lazy_mode mode)
>  
>  enum xen_lazy_mode xen_get_lazy_mode(void);
>  
> +#if defined(CONFIG_XEN_DOM0) && defined(CONFIG_ACPI)
> +void xen_sanitize_proc_cap_bits(uint32_t *buf);
> +#else
> +static inline void xen_sanitize_proc_cap_bits(uint32_t *buf)
> +{
> +	BUG();
> +}
> +#endif
> +
>  #endif /* _ASM_X86_XEN_HYPERVISOR_H */
> diff --git a/drivers/xen/pcpu.c b/drivers/xen/pcpu.c
> index b3e3d1bb37f3..7000701dff8f 100644
> --- a/drivers/xen/pcpu.c
> +++ b/drivers/xen/pcpu.c
> @@ -47,6 +47,9 @@
>  #include <asm/xen/hypervisor.h>
>  #include <asm/xen/hypercall.h>
>  
> +#ifdef CONFIG_ACPI
> +#include <acpi/processor.h>
> +#endif
>  
>  /*
>   * @cpu_id: Xen physical cpu logic number
> @@ -400,4 +403,22 @@ bool __init xen_processor_present(uint32_t acpi_id)
>  
>  	return online;
>  }
> +
> +void xen_sanitize_proc_cap_bits(uint32_t *cap)
> +{
> +	struct xen_platform_op op = {
> +		.cmd			= XENPF_set_processor_pminfo,
> +		.u.set_pminfo.id	= -1,
> +		.u.set_pminfo.type	= XEN_PM_PDC,
> +	};
> +	u32 buf[3] = { ACPI_PDC_REVISION_ID, 1, *cap };
> +	int ret;
> +
> +	set_xen_guest_handle(op.u.set_pminfo.pdc, buf);
> +	ret = HYPERVISOR_platform_op(&op);
> +	if (ret)
> +		pr_err("sanitize of _PDC buffer bits from Xen failed: %d\n",
> +		       ret);
> +	*cap = buf[2];

FWIW, we might want to only update cap if the hypercall has been
successful, otherwise there's no guarantee of what's in the buffer, so
I would recommend to put the updating of cap in an else branch.

Anyway, not a strong opinion, as I think in practice even if the
hypercall fails it wouldn't corrupt the data in the buffer, but seems
more robust.

Thanks, Roger.