Re: [PATCH v2] arm_mpam: resctrl: Limit num_rmids to the CLOSIDs in use

[Jinjie Ruan <ruanjinjie@huawei.com>]

在 2026/11/5 22:12, Ben Horgan 写道:
> Hi Jinjie,
> 
> For patches that touch mpam and fs/resctrl the conventional prefix, as I understand it, should be
> arm,fs/resctrl: and if it also touches the x86 rdt code it should then be arm,x86,fs/resctrl:.

Hi Ben,

Thank you for the reminder.

> Although, if possible please try and make separate patches for each.
> 
> 
> On 05/11/2026 12:33, Jinjie Ruan wrote:
>> On MPAM an RMID is a PMG within a PARTID, so resctrl can only allocate
>> the PMGs of the CLOSIDs it uses. info/L3_MON/num_rmids is instead set
>> to the whole PARTID and PMG space when the monitors are initialised,
>> and halved when CDP is enabled. It reports more RMIDs than resctrl can
>> allocate in two cases:
>>
>> - With CDP each CLOSID takes two PARTIDs, and resctrl halves the
>>   CLOSID count rounding down, while MPAM halves the whole RMID space.
>>   With 5 PARTIDs and 2 PMGs resctrl has 2 CLOSIDs and 4 RMIDs, but
>>   num_rmids reads 5. The count is too high whenever the number of
>>   PARTIDs is odd and there is more than one PMG.
>>
>> - On a monitor-only system there are no control groups and resctrl
>>   only uses the default CLOSID, but num_rmids still covers every
>>   PARTID. With 5 PARTIDs and 2 PMGs it reads 10 while only 2 RMIDs
>>   can be allocated.
>>
>> RMID allocation is limited to the CLOSIDs in use, so only the value
>> reported to userspace is too high. Creating as many monitor groups as
>> num_rmids allows fails with -ENOSPC before that number is reached.
>>
>> Let resctrl tell the architecture how many CLOSIDs it uses on every
>> mount. Add resctrl_arch_set_mon_rmid_count(), called after closid_init()
> 
> Is it really necessary for resctrl to tell the MPAM driver the number of CLOSIDs? Is there any case
> where the MPAM driver doesn't just know how many CLOSIDs there will be?

Unnecessary, we can do the same thing in the mpam driver, will drop it.

> 
> Thanks,
> 
> Ben
> 
> 
>> and before the info files are created, and have MPAM report
>> (pmg_max + 1) RMIDs for each CLOSID. The CLOSID count already accounts
>> for CDP, as schemata_list_add() halves it when CDP is enabled, so drop
>> the halving in resctrl_arch_set_cdp_enabled(). A monitor-only system
>> passes 1 for the default CLOSID. On x86 the RMIDs do not depend on the
>> CLOSIDs and the hook does nothing.
>>
>> Fixes: 264c285999fc ("arm_mpam: resctrl: Add monitor initialisation and domain boilerplate")
>> Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
>> Cc: <stable@vger.kernel.org> # 7.1.x
>> Cc: James Morse <james.morse@arm.com>
>> Cc: Ben Horgan <ben.horgan@arm.com>
>> Cc: Reinette Chatre <reinette.chatre@intel.com>
>> Cc: Fenghua Yu <fenghuay@nvidia.com>
>> ---
>> v2:
>> - Use the "arm_mpam: resctrl:" prefix.
>> - Also fix num_rmids on monitor-only systems, as Ben suggested: add
>>   resctrl_arch_set_mon_rmid_count() so resctrl reports the number of
>>   CLOSIDs it uses, and remove the CDP-only rounding.
>> - Retitle from "arm_mpam: Round the CDP RMID count down to whole CLOSIDs".
>> - v1: https://lore.kernel.org/all/20261104101622.563280-1-ruanjinjie@huawei.com/
>> ---
>>  arch/x86/include/asm/resctrl.h |  3 +++
>>  drivers/resctrl/mpam_resctrl.c | 18 ++++++++++++------
>>  fs/resctrl/rdtgroup.c          |  2 ++
>>  include/linux/resctrl.h        |  1 +
>>  4 files changed, 18 insertions(+), 6 deletions(-)
>>
>> diff --git a/arch/x86/include/asm/resctrl.h b/arch/x86/include/asm/resctrl.h
>> index 8f6edcdcfd87..50165027b92d 100644
>> --- a/arch/x86/include/asm/resctrl.h
>> +++ b/arch/x86/include/asm/resctrl.h
>> @@ -164,6 +164,9 @@ static inline bool resctrl_arch_match_rmid(struct task_struct *tsk, u32 ignored,
>>  	return READ_ONCE(tsk->rmid) == rmid;
>>  }
>>  
>> +/* RMIDs are independent of CLOSIDs on x86. */
>> +static inline void resctrl_arch_set_mon_rmid_count(u32 num_closid) { }
>> +
>>  static inline void resctrl_arch_sched_in(struct task_struct *tsk)
>>  {
>>  	if (static_branch_likely(&rdt_enable_key))
>> diff --git a/drivers/resctrl/mpam_resctrl.c b/drivers/resctrl/mpam_resctrl.c
>> index 5468367776e2..cca9568bab70 100644
>> --- a/drivers/resctrl/mpam_resctrl.c
>> +++ b/drivers/resctrl/mpam_resctrl.c
>> @@ -217,11 +217,6 @@ int resctrl_arch_set_cdp_enabled(enum resctrl_res_level rid, bool enable)
>>  	cdp_enabled = enable;
>>  	mpam_resctrl_controls[rid].cdp_enabled = enable;
>>  
>> -	if (enable)
>> -		l3->mon.num_rmid = resctrl_arch_system_num_rmid_idx() / 2;
>> -	else
>> -		l3->mon.num_rmid = resctrl_arch_system_num_rmid_idx();
>> -
>>  	/* The mbw_max feature can't hide cdp as it's a per-partid maximum. */
>>  	if (cdp_enabled && !mpam_resctrl_controls[RDT_RESOURCE_MBA].cdp_enabled)
>>  		mpam_resctrl_controls[RDT_RESOURCE_MBA].resctrl_res.alloc_capable = false;
>> @@ -279,6 +274,16 @@ u32 resctrl_arch_system_num_rmid_idx(void)
>>  	return (mpam_pmg_max + 1) * (mpam_partid_max + 1);
>>  }
>>  
>> +void resctrl_arch_set_mon_rmid_count(u32 num_closid)
>> +{
>> +	struct rdt_resource *l3 = &mpam_resctrl_controls[RDT_RESOURCE_L3].resctrl_res;
>> +
>> +	if (!l3->mon_capable)
>> +		return;
>> +
>> +	l3->mon.num_rmid = (mpam_pmg_max + 1) * num_closid;
>> +}
>> +
>>  u32 resctrl_arch_rmid_idx_encode(u32 closid, u32 rmid)
>>  {
>>  	return closid * (mpam_pmg_max + 1) + rmid;
>> @@ -1350,7 +1355,8 @@ static int mpam_resctrl_monitor_init(struct mpam_resctrl_mon *mon,
>>  	 * each control group. Hence, advertise the whole rmid_idx space even
>>  	 * though each control group has its own pmg/rmid space. Unfortunately,
>>  	 * this does mean userspace needs to know the architecture to correctly
>> -	 * interpret this value.
>> +	 * interpret this value. resctrl_arch_set_mon_rmid_count() replaces it
>> +	 * with the RMIDs of the CLOSIDs in use when resctrl is mounted.
>>  	 */
>>  	l3->mon.num_rmid = resctrl_arch_system_num_rmid_idx();
>>  
>> diff --git a/fs/resctrl/rdtgroup.c b/fs/resctrl/rdtgroup.c
>> index 106e78fb1f1d..2335dd7054ca 100644
>> --- a/fs/resctrl/rdtgroup.c
>> +++ b/fs/resctrl/rdtgroup.c
>> @@ -3214,6 +3214,8 @@ static int rdt_get_tree(struct fs_context *fc)
>>  	if (ret)
>>  		goto out_schemata_free;
>>  
>> +	resctrl_arch_set_mon_rmid_count(closids_supported() ?: 1);
>> +
>>  	if (resctrl_arch_mon_capable())
>>  		flags |= RFTYPE_MON;
>>  
>> diff --git a/include/linux/resctrl.h b/include/linux/resctrl.h
>> index 10dfdca7f4bf..4a78f2e60bdf 100644
>> --- a/include/linux/resctrl.h
>> +++ b/include/linux/resctrl.h
>> @@ -417,6 +417,7 @@ static inline u32 resctrl_get_default_ctrl(struct rdt_resource *r)
>>  /* The number of closid supported by this resource regardless of CDP */
>>  u32 resctrl_arch_get_num_closid(struct rdt_resource *r);
>>  u32 resctrl_arch_system_num_rmid_idx(void);
>> +void resctrl_arch_set_mon_rmid_count(u32 num_closid);
>>  int resctrl_arch_update_domains(struct rdt_resource *r, u32 closid);
>>  
>>  bool resctrl_enable_mon_event(enum resctrl_event_id eventid, bool any_cpu,
> 

-- 
Best regards,
Jinjie