Re: [PATCH] cpu-topology: warn if NUMA configurations conflicts with lower layer

[Dietmar Eggemann]

On 02/01/2020 14:22, Valentin Schneider wrote:
> On 02/01/2020 12:47, Zengtao (B) wrote:

[...]

>> From schedule view of point, lower level sched domain should be a subset of higher
>> Level sched domain.
>>
> 
> Right, and that is checked when you have sched_debug on the cmdline
> (or write 1 to /sys/kernel/debug/sched_debug & regenerate the sched domains)

You should even get informed in case CONFIG_SCHED_DEBUG is not set.

   BUG: arch topology borken

With CONFIG_SCHED_DEBUG (and a CPU removed from the cpu_mask (DIE level)
on an Arm64 Juno board) you get extra information:

   BUG: arch topology borken
        the MC domain not a subset of the DIE domain

> Now, I don't know how this plays out for the numa-in-package topologies like
> the one suggested by Sudeep. x86 and AMD had to play some games to get
> numa-in-package topologies working, see e.g.
> 
>   cebf15eb09a2 ("x86, sched: Add new topology for multi-NUMA-node CPUs")
> 

Yeah, the reason why we need this change would be interesting.

[...]