Re: [PATCH] fs/resctrl: Disallow monitor group rename when the RMID changes

[Jinjie Ruan <ruanjinjie@huawei.com>]

在 2026/11/4 22:36, Ben Horgan 写道:
> Hi Jinjie,
> 
> On 04/11/2026 10:19, Jinjie Ruan wrote:
>> From: Peter Newman <peternewman@google.com>
>>
>> Renaming a monitor group to move it below a different control group
>> changes the CLOSID of its tasks. Where the RMID only has meaning
>> together with the CLOSID, the RMID index changes as well: on arm64 MPAM
>> the RMID is a PMG that extends the PARTID, so the group's counters
>> belong to the old PARTID and cannot follow the move. The monitor data
>> would be reset and would then report the new PARTID's traffic.
>>
>> Check that the RMID index is unchanged before moving the group and
>> reject the rename with -EOPNOTSUPP otherwise. Only MPAM systems are
>> affected; architectures where the RMID is independent of the CLOSID,
>> such as x86, keep the same index and still allow the move. Document
>> the restriction in resctrl.rst.
>>
>> Fixes: 3e9b35823aab ("arm_mpam: resctrl: Add rmid index helpers")
>> Suggested-by: Ben Horgan <ben.horgan@arm.com>
> 
> Perhaps stretching the meaning of a 'Suggested-by' tag as I only suggested the repost of Peter's patch.
> 
> It is probably worth pointing out that the moving of monitor groups is existingly broken other than
> just the resetting of mbm_total_bytes counters and so the current behaviour isn't something that an
> existing user could be relying on. As you already point out in [1]

Sorry, will delete the Suggested-by and supplement the information in [1].

> 
> [1] https://lore.kernel.org/lkml/20261031101052.1692230-1-ruanjinjie@huawei.com/
> 
> Thanks,
> 
> Ben
> 
>> Signed-off-by: Peter Newman <peternewman@google.com>
>> Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
>> Cc: <stable@vger.kernel.org> # 7.1.x
>> Link: https://lore.kernel.org/lkml/20241205153845.394714-1-peternewman@google.com/
>> ---
>>  Documentation/filesystems/resctrl.rst |  7 ++++---
>>  fs/resctrl/rdtgroup.c                 | 13 +++++++++++++
>>  2 files changed, 17 insertions(+), 3 deletions(-)
>>
>> diff --git a/Documentation/filesystems/resctrl.rst b/Documentation/filesystems/resctrl.rst
>> index b52795e03303..5f21ab20030a 100644
>> --- a/Documentation/filesystems/resctrl.rst
>> +++ b/Documentation/filesystems/resctrl.rst
>> @@ -552,9 +552,10 @@ will automatically remove all MON groups below it.
>>  Moving MON group directories to a new parent CTRL_MON group is supported
>>  for the purpose of changing the resource allocations of a MON group
>>  without impacting its monitoring data or assigned tasks. This operation
>> -is not allowed for MON groups which monitor CPUs. No other move
>> -operation is currently allowed other than simply renaming a CTRL_MON or
>> -MON group.
>> +is not allowed for MON groups which monitor CPUs, and fails with
>> +-EOPNOTSUPP on systems where the RMID depends on the CLOSID, such as
>> +arm64 MPAM. No other move operation is currently allowed other than
>> +simply renaming a CTRL_MON or MON group.
>>  
>>  All groups contain the following files:
>>  
>> diff --git a/fs/resctrl/rdtgroup.c b/fs/resctrl/rdtgroup.c
>> index 9290a346386e..358f2dc8301d 100644
>> --- a/fs/resctrl/rdtgroup.c
>> +++ b/fs/resctrl/rdtgroup.c
>> @@ -4405,6 +4405,19 @@ static int rdtgroup_rename(struct kernfs_node *kn,
>>  		goto out;
>>  	}
>>  
>> +	/*
>> +	 * Moving the group changes the CLOSID. Where the RMID is only
>> +	 * meaningful together with the CLOSID, the RMID index changes and the
>> +	 * monitor state cannot follow the move.
>> +	 */
>> +	if (rdtgrp->mon.parent != new_prdtgrp && resctrl_arch_mon_capable() &&
>> +	    resctrl_arch_rmid_idx_encode(rdtgrp->closid, rdtgrp->mon.rmid) !=
>> +	    resctrl_arch_rmid_idx_encode(new_prdtgrp->closid, rdtgrp->mon.rmid)) {
>> +		rdt_last_cmd_puts("Changing parent control group not supported\n");
>> +		ret = -EOPNOTSUPP;
>> +		goto out;
>> +	}
>> +
>>  	/*
>>  	 * If the MON group is monitoring CPUs, the CPUs must be assigned to the
>>  	 * current parent CTRL_MON group and therefore cannot be assigned to
> 

-- 
Best regards,
Jinjie