Re: [PATCH v4 11/19] x86/resctrl: Introduce mbm_total_cfg and mbm_local_cfg

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

Hi Babu,

On 6/27/24 11:51 AM, Moger, Babu wrote:
> Hi Reinette,
> 
> On 6/13/24 20:43, Reinette Chatre wrote:
>> Hi Babu,
>>
>> On 5/24/24 5:23 AM, Babu Moger wrote:
>>> If the BMEC (Bandwidth Monitoring Event Configuration) feature is
>>> supported, the bandwidth events can be configured to track specific
>>> events. The event configuration is domain specific. ABMC (Assignable
>>> Bandwidth Monitoring Counters) feature needs event configuration
>>> information to assign hardware counter to an RMID. Event configurations
>>> are not stored in resctrl but instead always read from or written to
>>> hardware directly when prompted by user space.
>>>
>>> Read the event configuration from the hardware during the domain
>>> initialization. Save the configuration information in the rdt_hw_domain,
>>> so it can be used for counter assignment.
>>>
>>> Signed-off-by: Babu Moger <babu.moger@amd.com>
>>> ---
>>> v4: Read the configuration information from the hardware to initialize.
>>>       Added few commit messages.
>>>       Fixed the tab spaces.
>>>
>>> v3: Minor changes related to rebase in mbm_config_write_domain.
>>>
>>> v2: No changes.
>>> ---
>>>    arch/x86/kernel/cpu/resctrl/core.c     |  2 ++
>>>    arch/x86/kernel/cpu/resctrl/internal.h |  5 +++++
>>>    arch/x86/kernel/cpu/resctrl/monitor.c  | 21 +++++++++++++++++++++
>>>    3 files changed, 28 insertions(+)
>>>
>>> diff --git a/arch/x86/kernel/cpu/resctrl/core.c
>>> b/arch/x86/kernel/cpu/resctrl/core.c
>>> index ec93f6a50308..856c46d12177 100644
>>> --- a/arch/x86/kernel/cpu/resctrl/core.c
>>> +++ b/arch/x86/kernel/cpu/resctrl/core.c
>>> @@ -542,6 +542,8 @@ static void domain_add_cpu(int cpu, struct
>>> rdt_resource *r)
>>>            return;
>>>        }
>>>    +    arch_domain_mbm_evt_config(hw_dom);
>>> +
>>>        list_add_tail_rcu(&d->list, add_pos);
>>>          err = resctrl_online_domain(r, d);
>>> diff --git a/arch/x86/kernel/cpu/resctrl/internal.h
>>> b/arch/x86/kernel/cpu/resctrl/internal.h
>>> index 5e7e76cd512f..60a1ca0a11a7 100644
>>> --- a/arch/x86/kernel/cpu/resctrl/internal.h
>>> +++ b/arch/x86/kernel/cpu/resctrl/internal.h
>>> @@ -373,6 +373,8 @@ struct arch_mbm_state {
>>>     * @ctrl_val:    array of cache or mem ctrl values (indexed by CLOSID)
>>>     * @arch_mbm_total:    arch private state for MBM total bandwidth
>>>     * @arch_mbm_local:    arch private state for MBM local bandwidth
>>> + * @mbm_total_cfg:    MBM total bandwidth configuration
>>> + * @mbm_local_cfg:    MBM local bandwidth configuration
>>>     *
>>>     * Members of this structure are accessed via helpers that provide
>>> abstraction.
>>>     */
>>> @@ -381,6 +383,8 @@ struct rdt_hw_domain {
>>>        u32                *ctrl_val;
>>>        struct arch_mbm_state        *arch_mbm_total;
>>>        struct arch_mbm_state        *arch_mbm_local;
>>> +    u32                mbm_total_cfg;
>>> +    u32                mbm_local_cfg;
>>>    };
>>
>> Similar to the abmc_enabled member of rdt_hw_resource, these new
>> members of rdt_hw_domain are architecture specific and should never be
>> touched directly by resctrl fs code, for example, from mbm_config_show().
> 
> Need some clarification here.
> 
> I am thinking you want to introduce architecture specific routines to get
> and set mbm_total_config/mbm_local_config for the domain.
> Something like this.
> 
> +int arch_get_mbm_evt_cfg(struct rdt_domain *d, enum resctrl_event_id eventid)

The prefix for arch specific calls converged to "resctrl_arch_".

> +{
> +       struct rdt_hw_domain *hw_dom = resctrl_to_arch_dom(d);
> +
> +       switch (eventid) {
> +       case QOS_L3_OCCUP_EVENT_ID:
> +               break;
> +       case QOS_L3_MBM_TOTAL_EVENT_ID:
> +               return hw_dom->mbm_total_cfg;
> +       case QOS_L3_MBM_LOCAL_EVENT_ID:
> +               return hw_dom->mbm_local_cfg;
> +       }
> +
> +       /* Never expect to get here */
> +       WARN_ON_ONCE(1);
> +
> +       return -1;
> +}
> +
> +void arch_set_mbm_evt_cfg(struct rdt_domain *d,
> +                         enum resctrl_event_id eventid, u32 val)
> +{
> +       struct rdt_hw_domain *hw_dom = resctrl_to_arch_dom(d);
> +
> +       switch (eventid) {
> +       case QOS_L3_OCCUP_EVENT_ID:
> +               break;
> +       case QOS_L3_MBM_TOTAL_EVENT_ID:
> +               hw_dom->mbm_total_cfg = val;
> +               break;
> +       case QOS_L3_MBM_LOCAL_EVENT_ID:
> +               hw_dom->mbm_local_cfg = val;
> +       }
> +
> +       return;
> +}
> +

I expected that a call to set the event configuration on an architecture to
also interact with the hardware. Essentially what mon_event_config_write()
does today, but in addition also sets the cached mbm_total_cfg/mbm_local_cfg.

Part of this is done by the new MPAM portion [1]. With that the post-ABMC
implementation of resctrl_arch_mon_event_config_read() may be what you
have as arch_get_mbm_evt_cfg() above and resctrl_arch_mon_event_config_write()
does the same as in [1] but with addition of updating the cached
mbm_local_cfg/mbm_total_cfg within struct rdt_hw_domain. Would this work
for ABMC?
  
Reinette

[1] https://lore.kernel.org/lkml/20240614150033.10454-21-james.morse@arm.com/