linux-kernel - RE: [PATCH v8] cpuidle, ACPI: Evaluate LPI arch

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CH0PR02MB8073FBB64A196EEEFED11BB7F6C5A@CH0PR02MB8073.namprd02.prod.outlook.com>
Date:   Mon, 2 Oct 2023 19:22:57 +0000
From:   "Pawandeep Oza (QUIC)" <quic_poza@...cinc.com>
To:     'Will Deacon' <will@...nel.org>,
        "Pawandeep Oza (QUIC)" <quic_poza@...cinc.com>
CC:     "sudeep.holla@....com" <sudeep.holla@....com>,
        "catalin.marinas@....com" <catalin.marinas@....com>,
        "rafael@...nel.org" <rafael@...nel.org>,
        "lenb@...nel.org" <lenb@...nel.org>,
        "linux-arm-kernel@...ts.infradead.org" 
        <linux-arm-kernel@...ts.infradead.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-acpi@...r.kernel.org" <linux-acpi@...r.kernel.org>
Subject: RE: [PATCH v8] cpuidle, ACPI: Evaluate LPI arch_flags for broadcast
 timer



-----Original Message-----
From: Will Deacon <will@...nel.org> 
Sent: Friday, September 29, 2023 8:05 AM
To: Pawandeep Oza (QUIC) <quic_poza@...cinc.com>
Cc: sudeep.holla@....com; catalin.marinas@....com; rafael@...nel.org; lenb@...nel.org; linux-arm-kernel@...ts.infradead.org; linux-kernel@...r.kernel.org; linux-acpi@...r.kernel.org
Subject: Re: [PATCH v8] cpuidle, ACPI: Evaluate LPI arch_flags for broadcast timer

On Mon, Sep 18, 2023 at 10:21:40AM -0700, Oza Pawandeep wrote:
> Arm(r) Functional Fixed Hardware Specification defines LPI states, which 
> provide an architectural context loss flags field that can be used to 
> describe the context that might be lost when an LPI state is entered.
> 
> - Core context Lost
>         - General purpose registers.
>         - Floating point and SIMD registers.
>         - System registers, include the System register based
>         - generic timer for the core.
>         - Debug register in the core power domain.
>         - PMU registers in the core power domain.
>         - Trace register in the core power domain.
> - Trace context loss
> - GICR
> - GICD
> 
> Qualcomm's custom CPUs preserves the architectural state, including 
> keeping the power domain for local timers active.
> when core is power gated, the local timers are sufficient to wake the 
> core up without needing broadcast timer.
> 
> The patch fixes the evaluation of cpuidle arch_flags, and moves only 
> to broadcast timer if core context lost is defined in ACPI LPI.
> 
> Fixes: a36a7fecfe607 ("Add support for Low Power Idle(LPI) states")
> Reviewed-by: Sudeep Holla <sudeep.holla@....com>
> Acked-by: Rafael J. Wysocki <rafael@...nel.org>
> Signed-off-by: Oza Pawandeep <quic_poza@...cinc.com>
> ---
> 
> Notes:
>     Will/Catalin: Rafael has acked and he prefers to take it via arm64 
> tree
> 
> diff --git a/arch/arm64/include/asm/acpi.h 
> b/arch/arm64/include/asm/acpi.h index 4d537d56eb84..269d21209723 
> 100644
> --- a/arch/arm64/include/asm/acpi.h
> +++ b/arch/arm64/include/asm/acpi.h
> @@ -9,6 +9,7 @@
>  #ifndef _ASM_ACPI_H
>  #define _ASM_ACPI_H
>  
> +#include <linux/cpuidle.h>
>  #include <linux/efi.h>
>  #include <linux/memblock.h>
>  #include <linux/psci.h>
> @@ -44,6 +45,23 @@
>  
>  #define ACPI_MADT_GICC_TRBE  (offsetof(struct acpi_madt_generic_interrupt, \
>  	trbe_interrupt) + sizeof(u16))
> +/*
> + * Arm(r) Functional Fixed Hardware Specification Version 1.2.
> + * Table 2: Arm Architecture context loss flags  */
> +#define CPUIDLE_CORE_CTXT		BIT(0) /* Core context Lost */
> +
> +static __always_inline void _arch_update_idle_state_flags(u32 arch_flags,
> +							unsigned int *sflags)

Why can't this just be 'static inline'?

Oza: sure, will let compiler decide.

> +{
> +	if (arch_flags & CPUIDLE_CORE_CTXT)
> +		*sflags |= CPUIDLE_FLAG_TIMER_STOP; } #define 
> +arch_update_idle_state_flags _arch_update_idle_state_flags

Usually, the function and the macro have the same name for this pattern, so I think it would be more consistent to drop the leading underscore from the C function name.

Oza: sure

> +
> +#define CPUIDLE_TRACE_CTXT		BIT(1) /* Trace context loss */
> +#define CPUIDLE_GICR_CTXT		BIT(2) /* GICR */
> +#define CPUIDLE_GICD_CTXT		BIT(3) /* GICD */
>  
>  /* Basic configuration for ACPI */
>  #ifdef	CONFIG_ACPI
> diff --git a/drivers/acpi/processor_idle.c 
> b/drivers/acpi/processor_idle.c index dc615ef6550a..5c1d13eecdd1 
> 100644
> --- a/drivers/acpi/processor_idle.c
> +++ b/drivers/acpi/processor_idle.c
> @@ -1217,8 +1217,7 @@ static int acpi_processor_setup_lpi_states(struct acpi_processor *pr)
>  		strscpy(state->desc, lpi->desc, CPUIDLE_DESC_LEN);
>  		state->exit_latency = lpi->wake_latency;
>  		state->target_residency = lpi->min_residency;
> -		if (lpi->arch_flags)
> -			state->flags |= CPUIDLE_FLAG_TIMER_STOP;
> +		arch_update_idle_state_flags(lpi->arch_flags, &state->flags);

Hmm, I know Rafael has Acked this, but I think this is pretending to be more generic than it really is. While passing in a pointer to the flags field allows the arch code to set and clear arbitrary flags, we're calling this before we've set CPUIDLE_FLAG_RCU_IDLE, so that cannot be changed.

Why not just name it like it is and return the arch flags directly:

	state->flags |= arch_get_idle_state_flags(lpi->arch_flags);

Oza: 

?

>  		if (i != 0 && lpi->entry_method == ACPI_CSTATE_FFH)
>  			state->flags |= CPUIDLE_FLAG_RCU_IDLE;
>  		state->enter = acpi_idle_lpi_enter; diff --git 
> a/include/linux/acpi.h b/include/linux/acpi.h index 
> a73246c3c35e..07a825c76bab 100644
> --- a/include/linux/acpi.h
> +++ b/include/linux/acpi.h
> @@ -1480,6 +1480,12 @@ static inline int 
> lpit_read_residency_count_address(u64 *address)  }  #endif
>  
> +#ifdef CONFIG_ACPI_PROCESSOR_IDLE
> +#ifndef arch_update_idle_state_flags
> +#define arch_update_idle_state_flags(af, sf)	do {} while (0)

I'd prefer defining this to point at an empty static inline function so that we get evaluation and type-checking of the arguments.

Oza: sure

> +#endif
> +#endif /* CONFIG_ACPI_PROCESSOR_IDLE */

Why do you need the outer CONFIG_ guards here?

Oza: this is because of non-ACPI kernel build issue for this config: https://download.01.org/0day-ci/archive/20230915/202309151138.69mFCPtW-lkp@intel.com/config
Throwing following

" 
All warnings (new ones prefixed by >>):

   In file included from arch/arm64/kernel/setup.c:36:
>> arch/arm64/include/asm/acpi.h:60: warning: 
>> "arch_update_idle_state_flags" redefined
      60 | #define arch_update_idle_state_flags _arch_update_idle_state_flags
         | 
   In file included from arch/arm64/kernel/setup.c:9:
   include/linux/acpi.h:1484: note: this is the location of the previous definition
    1484 | #define arch_update_idle_state_flags(af, sf)    do {} while (0)
         |
"

Will