Re: [PATCH 1/2] x86/resctrl, Documentation: Keep mbm_assign_mode "default" on boot
Reinette Chatre <[email protected]> Tue, 4 Aug 2026 17:12:57 -0700
| Newsgroups | gmane.linux.documentation,gmane.linux.kernel |
|---|---|
| Message-ID | <[email protected]> |
On 8/4/26 4:35 PM, Reinette Chatre wrote: > On 8/4/26 3:47 PM, Borislav Petkov wrote: >> With my simple thinking, I would expect that accurate data means, the number >> of counters being in use is not hitting the arch limit. The moment that >> happens, I guess one could deem that measurement innacurate. > > Agreed and matches above summary. The arch limit here would be the 64 RMID that can > be guaranteed to be counted. resctrl does not limit the number of monitor groups to > this number though but instead uses the number enumerated from hardware (4096 on this hardware). > To guarantee accurate data in "default" mode a solution could be to add model specific > information that teaches resctrl about "64" and it can use that as RMID limit instead. As a follow-up this is not as simple since at this time this limit would also be applied to assignable counter mode, which is not intended. Another caveat is this will present significantly fewer RMIDs than what users are used to. The current argument as I see it is that users are used to many RMID with inaccurate data so this patch aims to keep giving them inaccurate data even though hardware is now capable of providing accurate data. Reinette