Re: [PATCH v5 13/29] x86,fs/resctrl: Handle events that can be read from any CPU

[Reinette Chatre <reinette.chatre@intel.com>]

Hi Tony,

On 5/21/25 3:50 PM, Tony Luck wrote:
> Resctrl file system code was built with the assumption that monitor
> events can only be read from a CPU in the cpumask_t set for each
> domain.
> 
> This was true for x86 events accessed with an MSR interface, but may
> not be true for other access methods such as MMIO.
> 
> Add a flag to struct mon_evt to indicate which events can be read on
> any CPU.

Since struct mon_evt is per-event, how about:
"Add a flag to struct mon_evt to indicate if the event can be read on
any CPU."

> 
> Architecture uses resctrl_enable_mon_event() to enable an event and
> set the flag appropriately.
> 
> Bypass all the smp_call*() code for events that can be read on any CPU
> and call mon_event_count() directly from mon_event_read().
> 
> Skip checks in __mon_event_count() that the read is being done from
> a CPU in the correct domain or cache scope.
> 
> Signed-off-by: Tony Luck <tony.luck@intel.com>
> ---
>  include/linux/resctrl.h            |  2 +-
>  fs/resctrl/internal.h              |  2 ++
>  arch/x86/kernel/cpu/resctrl/core.c |  6 +++---
>  fs/resctrl/ctrlmondata.c           |  7 ++++++-
>  fs/resctrl/monitor.c               | 26 ++++++++++++++++++++------
>  5 files changed, 32 insertions(+), 11 deletions(-)
> 
> diff --git a/include/linux/resctrl.h b/include/linux/resctrl.h
> index b7a4c7bf4feb..9aab3d78005a 100644
> --- a/include/linux/resctrl.h
> +++ b/include/linux/resctrl.h
> @@ -377,7 +377,7 @@ u32 resctrl_arch_get_num_closid(struct rdt_resource *r);
>  u32 resctrl_arch_system_num_rmid_idx(void);
>  int resctrl_arch_update_domains(struct rdt_resource *r, u32 closid);
>  
> -void resctrl_enable_mon_event(enum resctrl_event_id evtid);
> +void resctrl_enable_mon_event(enum resctrl_event_id evtid, bool any_cpu);
>  
>  bool resctrl_is_mon_event_enabled(enum resctrl_event_id evt);
>  
> diff --git a/fs/resctrl/internal.h b/fs/resctrl/internal.h
> index 085a2ee1922f..eb6e92d1ab15 100644
> --- a/fs/resctrl/internal.h
> +++ b/fs/resctrl/internal.h
> @@ -57,6 +57,7 @@ static inline struct rdt_fs_context *rdt_fc2context(struct fs_context *fc)
>   * @rid:		index of the resource for this event
>   * @name:		name of the event
>   * @configurable:	true if the event is configurable
> + * @any_cpu:		true if the event can be read from any CPU
>   * @enabled:		true if the event is enabled
>   */
>  struct mon_evt {
> @@ -64,6 +65,7 @@ struct mon_evt {
>  	enum resctrl_res_level	rid;
>  	char			*name;
>  	bool			configurable;
> +	bool			any_cpu;
>  	bool			enabled;
>  };
>  
> diff --git a/arch/x86/kernel/cpu/resctrl/core.c b/arch/x86/kernel/cpu/resctrl/core.c
> index b39537658618..5d9a024ce4b0 100644
> --- a/arch/x86/kernel/cpu/resctrl/core.c
> +++ b/arch/x86/kernel/cpu/resctrl/core.c
> @@ -880,15 +880,15 @@ static __init bool get_rdt_mon_resources(void)
>  	bool ret = false;
>  
>  	if (rdt_cpu_has(X86_FEATURE_CQM_OCCUP_LLC)) {
> -		resctrl_enable_mon_event(QOS_L3_OCCUP_EVENT_ID);
> +		resctrl_enable_mon_event(QOS_L3_OCCUP_EVENT_ID, false);
>  		ret = true;
>  	}
>  	if (rdt_cpu_has(X86_FEATURE_CQM_MBM_TOTAL)) {
> -		resctrl_enable_mon_event(QOS_L3_MBM_TOTAL_EVENT_ID);
> +		resctrl_enable_mon_event(QOS_L3_MBM_TOTAL_EVENT_ID, false);
>  		ret = true;
>  	}
>  	if (rdt_cpu_has(X86_FEATURE_CQM_MBM_LOCAL)) {
> -		resctrl_enable_mon_event(QOS_L3_MBM_LOCAL_EVENT_ID);
> +		resctrl_enable_mon_event(QOS_L3_MBM_LOCAL_EVENT_ID, false);
>  		ret = true;
>  	}
>  
> diff --git a/fs/resctrl/ctrlmondata.c b/fs/resctrl/ctrlmondata.c
> index dcde27f6f2ec..1337716f59c8 100644
> --- a/fs/resctrl/ctrlmondata.c
> +++ b/fs/resctrl/ctrlmondata.c
> @@ -569,6 +569,11 @@ void mon_event_read(struct rmid_read *rr, struct rdt_resource *r,
>  		return;
>  	}
>  
> +	if (evt->any_cpu) {
> +		mon_event_count(rr);
> +		goto done;

Please see earlier details about goto in resctrl. This can be "out_ctx_free".

> +	}
> +
>  	cpu = cpumask_any_housekeeping(cpumask, RESCTRL_PICK_ANY_CPU);
>  
>  	/*
> @@ -581,7 +586,7 @@ void mon_event_read(struct rmid_read *rr, struct rdt_resource *r,
>  		smp_call_function_any(cpumask, mon_event_count, rr, 1);
>  	else
>  		smp_call_on_cpu(cpu, smp_mon_event_count, rr, false);
> -
> +done:
>  	resctrl_arch_mon_ctx_free(r, evt->evtid, rr->arch_mon_ctx);
>  }
>  
> diff --git a/fs/resctrl/monitor.c b/fs/resctrl/monitor.c
> index 3cfd1bf1845e..e6e3be990638 100644
> --- a/fs/resctrl/monitor.c
> +++ b/fs/resctrl/monitor.c
> @@ -356,9 +356,24 @@ static struct mbm_state *get_mbm_state(struct rdt_l3_mon_domain *d, u32 closid,
>  	return states ? &states[idx] : NULL;
>  }
>  
> +static bool cpu_on_wrong_domain(struct rmid_read *rr)
> +{
> +	cpumask_t *mask;
> +
> +	if (rr->evt->any_cpu)
> +		return false;
> +
> +	/*
> +	 * When reading from a specific domain the CPU must be in that
> +	 * domain. Otherwise the CPU must be one that shares the cache.
> +	 */
> +	mask = rr->d ? &rr->d->hdr.cpu_mask : &rr->ci->shared_cpu_map;
> +
> +	return !cpumask_test_cpu(smp_processor_id(), mask);
> +}

I find double-negatives can trip people up. Having function name contain
"wrong" while also returning "!" can be confusing. I think this will be simpler
if it is a straight-forward utility, for example, "cpu_on_correct_domain()"?
Maybe even "current_cpu_on_correct_domain()" to be explicit about which CPU
is being checked.

> +
>  static int __mon_event_count(u32 closid, u32 rmid, struct rmid_read *rr)
>  {
> -	int cpu = smp_processor_id();
>  	struct rdt_l3_mon_domain *d;
>  	struct mbm_state *m;
>  	int err, ret;
> @@ -373,8 +388,7 @@ static int __mon_event_count(u32 closid, u32 rmid, struct rmid_read *rr)
>  	}
>  
>  	if (rr->d) {
> -		/* Reading a single domain, must be on a CPU in that domain. */
> -		if (!cpumask_test_cpu(cpu, &rr->d->hdr.cpu_mask))
> +		if (cpu_on_wrong_domain(rr))
>  			return -EINVAL;
>  		rr->err = resctrl_arch_rmid_read(rr->r, rr->d, closid, rmid,
>  						 rr->evt->evtid, &tval, rr->arch_mon_ctx);
> @@ -386,8 +400,7 @@ static int __mon_event_count(u32 closid, u32 rmid, struct rmid_read *rr)
>  		return 0;
>  	}
>  
> -	/* Summing domains that share a cache, must be on a CPU for that cache. */
> -	if (!cpumask_test_cpu(cpu, &rr->ci->shared_cpu_map))
> +	if (cpu_on_wrong_domain(rr))
>  		return -EINVAL;
>  
>  	/*
> @@ -865,7 +878,7 @@ struct mon_evt mon_event_all[QOS_NUM_EVENTS] = {
>  	},
>  };
>  
> -void resctrl_enable_mon_event(enum resctrl_event_id evtid)
> +void resctrl_enable_mon_event(enum resctrl_event_id evtid, bool any_cpu)
>  {
>  	if (WARN_ON_ONCE(evtid >= QOS_NUM_EVENTS))
>  		return;
> @@ -874,6 +887,7 @@ void resctrl_enable_mon_event(enum resctrl_event_id evtid)
>  		return;
>  	}
>  
> +	mon_event_all[evtid].any_cpu = any_cpu;
>  	mon_event_all[evtid].enabled = true;
>  }
>  

Reinette