Re: [PATCH v10 00/12] Introduce MMIO-based CMT access for Enhanced RDT

From: Chen Yu

Date: Mon Nov 16 2026 - 04:56:28 EST


On Thu, Nov 05, 2026 at 06:07:59PM +0800, Chen Yu wrote:
> Date: Thu, 5 Nov 2026 18:07:59 +0800
> From: Chen Yu <yu.c.chen@xxxxxxxxx>
> To: tony.luck@xxxxxxxxx, reinette.chatre@xxxxxxxxx
> Cc: tglx@xxxxxxxxxx, bp@xxxxxxxxx, mingo@xxxxxxxxxx,
> dave.hansen@xxxxxxxxxxxxxxx, hpa@xxxxxxxxx, fenghuay@xxxxxxxxxx,
> babu.moger@xxxxxxx, hongyu.ning@xxxxxxxxx, chen.yu@xxxxxxxxx,
> x86@xxxxxxxxxx, linux-kernel@xxxxxxxxxxxxxxx
> Subject: [PATCH v10 00/12] Introduce MMIO-based CMT access for Enhanced RDT
> X-Mailer: git-send-email 2.25.1
>
> v9: https://lore.kernel.org/lkml/cover.1791933001.git.yu.c.chen@xxxxxxxxx/
> v8: https://lore.kernel.org/lkml/cover.1789705667.git.yu.c.chen@xxxxxxxxx/
> v7: https://lore.kernel.org/lkml/cover.1787976868.git.yu.c.chen@xxxxxxxxx/
> v6: https://lore.kernel.org/lkml/cover.1784968626.git.yu.c.chen@xxxxxxxxx/
> v5: https://lore.kernel.org/lkml/cover.1782866200.git.yu.c.chen@xxxxxxxxx/
> v4: https://lore.kernel.org/lkml/cover.1781332698.git.yu.c.chen@xxxxxxxxx/
> v3: https://lore.kernel.org/lkml/cover.1780710620.git.yu.c.chen@xxxxxxxxx/
> v2: https://lore.kernel.org/lkml/cover.1780587063.git.yu.c.chen@xxxxxxxxx/
> v1: https://lore.kernel.org/lkml/cover.1779872016.git.yu.c.chen@xxxxxxxxx/
>
> Intel Enhanced Resource Director Technology (ERDT) extends the existing
> RDT framework with two major capabilities:
>
> 1. MMIO-based access to monitoring and allocation registers, replacing
> the legacy MSR-based interface.
> 2. Region-aware RDT for fine-grained control over different tiers of
> memory (e.g., CXL.mem, DDR).
>
> This is described in the Intel RDT Architecture Specification:
> https://cdrdv2-public.intel.com/789566/356688-intel-rdt-arch-spec.pdf
>
> This patch set focuses on the first part: enabling MMIO-based access for
> Cache Monitoring Technology (CMT), while CAT/MBM/MBA are still using MSR.
> The platform advertises the MMIO register layout through the ACPI ERDT
> (Enhanced Resource Director Technology) table, which contains sub-tables
> describing per-domain register regions for monitoring and allocation.
>
> With ERDT, L3 cache occupancy counters are read via MMIO rather than
> MSR, allowing the reads to be performed from any CPU without requiring
> cross-CPU IPIs. This series parses the relevant ACPI sub-tables (RMDD,
> CACD, CMRC), prepares the resctrl monitor infrastructure for MMIO-based
> reads, and adds initial support for reading L3 occupancy via the CMRC
> interface. ERDT is not compatible with Sub-NUMA Cluster (SNC) mode and
> stays disabled when SNC is enabled.
>
> Thanks Tony, Reinette, Thomas, Dave, Boris, Peter, Hongyu for your time to
> look at this patch set.
>
> Internal sashiko has not reported any issues on the v10. The change history
> in the section "Changes since v9" in each patch is generated by AI.
>

Response to the sashiko 2 review feedback at
https://sashiko.dev/#/patchset/cover.1793864720.git.yu.c.chen%40intel.com
(false positive IMO)

Review1:
"In resctrl_arch_pre_mount(), all online CPUs are then added to
RDT_RESOURCE_PERF_PKG without checking erdt_try_bind_domain():
arch/x86/kernel/cpu/resctrl/core.c:resctrl_arch_pre_mount() {
...
cpus_read_lock();
mutex_lock(&domain_list_lock);
r->mon_capable = true;
rdt_mon_capable = true;
for_each_online_cpu(cpu)
domain_add_cpu_mon(cpu, r);
mutex_unlock(&domain_list_lock);
cpus_read_unlock();
}
Will this cause an online CPU that failed erdt_try_bind_domain() at boot to
still be added to the RDT_RESOURCE_PERF_PKG monitoring domain on first mount,
while a CPU onlined (or offlined and re-onlined) after mount is excluded from
RDT_RESOURCE_PERF_PKG by resctrl_arch_online_cpu()?
Should resctrl_arch_pre_mount() also check erdt_try_bind_domain(cpu) before
calling domain_add_cpu_mon(cpu, r)?"

Response:
PERF_PKG(AET) does not care about ERDT information, so whether to add the "broken" CPU into
AET's domain does not impact the result - the AET monitor event's any_cpu is true, so
any CPU can read the monitor data without having to be read from the CPU in that domain.
We can keep as it is for now.

Review2:
Should cmrc_init() in arch/x86/kernel/cpu/resctrl/erdt.c also reject a zero
cmrc->up_scale (for example, !cmrc->up_scale || cmrc->up_scale > UINT_MAX)?

Response:
The correctness of cmrc->up_scale cannot be guaranteed, because we cannot distinguish between
a 0 value up_scale and any value between 0 and UINT_MAX, they have the same effect, and we have
to trust the BIOS/firmware on this up_scale. We can add the above check but are not sure if it is usable.

thanks,
Chenyu