On Thu, Jul 15, 2021 at 11:04:19AM +0200, Peter Zijlstra wrote:

On Thu, Jul 15, 2021 at 01:13:38AM +0200, Frederic Weisbecker wrote:

quoted

On Wed, Jul 14, 2021 at 06:52:43PM +0200, Peter Zijlstra wrote:

quoted

quoted

cpusets already has means to create paritions; why are you creating
something else?

I was about to answer that the semantics of isolcpus, which reference
a NULL domain, are different from SD_LOAD_BALANCE implied by
cpuset.sched_load_balance. But then I realize that SD_LOAD_BALANCE has
been removed.

How cpuset.sched_load_balance is implemented then? Commit
e669ac8ab952df2f07dee1e1efbf40647d6de332 ("sched: Remove checks against
SD_LOAD_BALANCE") advertize that setting cpuset.sched_load_balance to 0
ends up creating NULL domain but that's not what I get. For example if I
mount a single cpuset root (no other cpuset mountpoints):

SD_LOAD_BALANCE was only for when you wanted to stop balancing inside a
domain tree. That no longer happens (and hasn't for a *long* time).
Cpusets simply creates multiple domain trees (or the empty one if its
just one CPU).

Ok.

quoted

$ mount -t cgroup none ./cpuset -o cpuset
$ cd cpuset
$ cat cpuset.cpus
0-7
$ cat cpuset.sched_load_balance
1
$ echo 0 > cpuset.sched_load_balance
$ ls /sys/kernel/debug/domains/cpu1/
domain0  domain1

I still get the domains on all CPUs...

(note, that's the cgroup-v1 interface, the cgroup-v2 interface is
significantly different)

I'd suggest doing: echo 1 > /debug/sched/verbose, if I do the above I
get:

[1290784.889705] CPU0 attaching NULL sched-domain.
[1290784.894830] CPU1 attaching NULL sched-domain.

Thanks!

Eventually I uninstalled cgmanager and things seem to work now. I have
no idea why and I'm not sure I'm willing to investigate further :o)