[PATCH v4 38/41] arm_mpam: Add workaround for T241-MPAM-4

[Shaopeng Tan (Fujitsu)]

Hi Zengheng

> Tan, Shaopeng/譚 紹鵬;​Ben Horgan <ben.horgan at arm.com>​
> ​amitsinght at marvell.com;​baisheng.gao at unisoc.com;​baolin.wang at linux.alibaba.com;​carl at os.amperecomputing.com;​dave.martin at arm.com;​david at kernel.org;​dfustini at baylibre.com;​fenghuay at nvidia.com;​gshan at redhat.com;​james.morse at arm.com;​jonathan.cameron at huawei.com;​kobak at nvidia.com;​lcherian at marvell.com;​linux-arm-kernel at lists.infradead.org;​linux-kernel at vger.kernel.org;​他 +18 件​​​​​
> Hi Shaopeng,
> 
> On 2026/2/13 15:02, Shaopeng Tan (Fujitsu) wrote:
> > Hello Ben, Fenghua
> >
> >> From: Shanker Donthineni <sdonthineni at nvidia.com>
> >>
> >> In the T241 implementation of memory-bandwidth partitioning, in the absence
> >> of contention for bandwidth, the minimum bandwidth setting can affect the
> >> amount of achieved bandwidth. Specifically, the achieved bandwidth in the
> >> absence of contention can settle to any value between the values of
> >> MPAMCFG_MBW_MIN and MPAMCFG_MBW_MAX.  Also, if MPAMCFG_MBW_MIN is set
> >> zero (below 0.78125%), once a core enters a throttled state, it will never
> >> leave that state.
> >>
> >> The first issue is not a concern if the MPAM software allows to program
> >> MPAMCFG_MBW_MIN through the sysfs interface. This patch ensures program
> >> MBW_MIN=1 (0.78125%) whenever MPAMCFG_MBW_MIN=0 is programmed.
> >>
> >> In the scenario where the resctrl doesn't support the MBW_MIN interface via
> >> sysfs, to achieve bandwidth closer to MBW_MAX in the absence of contention,
> >> software should configure a relatively narrow gap between MBW_MIN and
> >> MBW_MAX. The recommendation is to use a 5% gap to mitigate the problem.
> >
> > I have a question regarding the MBW_MIN values.
> > Are there any cases where the sum of all MBW_MIN values from different control groups exceeds 100%?
> > And if so, is it acceptable for it to exceed 100%?"
> >
> > Best regards,
> > Shaopeng TAN
> >
> 
> 
> Per the ARM MPAM architecture specification: "A PARTID that has used
> less than MIN is given preferential access to bandwidth."
> 
> MBW_MIN is not a guaranteed bandwidth allocation. Instead, it serves as
> a priority threshold: when a partid's memory bandwidth usage falls below
> the configured MBW_MIN value, its priority for memory bandwidth access
> is elevated.
> 
> Therefore, it is acceptable for the sum of MBW_MIN values across
> different control groups to exceed 100%. There is no requirement
> for these values to add up to 100% or less.
> 
> Hope this clarifies the behavior.

I see, thank you.

Best regards,
Shaopeng TAN