Raspberry PI 3b+ offline CPUs

[mordax7]

I have an performance issues with my Raspberry PI 3b+ because of offline CPUs.

My setup:

• Hardware: Raspberry PI 3b+
• Image: kairos-alpine-arm-rpi-v1.1.6-k3sv1.25.3+k3s1.img
• cloud-config.yml:

#node-config
install:
  # Device for automated installs
  device: /dev/sda
  # Reboot after installation
  reboot: true
stages:
  initramfs:
    - name: User settings and hostname
      hostname: kairos-controller
      users:
        kairos:
          passwd: DEDUCTED
  network:
    - name: User settings
      authorized_keys:
        kairos:
          - github:mordax7
k3s:
  enabled: true
  args:
    - --cluster-init
    - --etcd-snapshot-schedule-cron 0 */12 * * *
    - --etcd-snapshot-retention 14
    - --etcd-snapshot-dir /var/lib/rancher/k3s/server/db/snapshots/
  env:
    K3S_TOKEN: DEDUCTED

When I use lscpu I see that 3 of my 4 CPUs are offline for some reason.:

lscpu
Architecture:            aarch64
  CPU op-mode(s):        32-bit, 64-bit
  Byte Order:            Little Endian
CPU(s):                  4
  On-line CPU(s) list:   0
  Off-line CPU(s) list:  1-3
Vendor ID:               ARM
  BIOS Vendor ID:        Unknown
  Model name:            Cortex-A53
    BIOS Model name:     Unknown  CPU @ 0.0GHz
    BIOS CPU family:     2
    Model:               4
    Thread(s) per core:  1
    Core(s) per cluster: 1
    Socket(s):           1
    Cluster(s):          1
    Stepping:            r0p4
    CPU(s) scaling MHz:  100%
    CPU max MHz:         1400.0000
    CPU min MHz:         600.0000
    BogoMIPS:            38.40
    Flags:               fp asimd evtstrm crc32 cpuid
NUMA:
  NUMA node(s):          1
  NUMA node0 CPU(s):     0
Vulnerabilities:
  Itlb multihit:         Not affected
  L1tf:                  Not affected
  Mds:                   Not affected
  Meltdown:              Not affected
  Mmio stale data:       Not affected
  Retbleed:              Not affected
  Spec store bypass:     Not affected
  Spectre v1:            Mitigation; __user pointer sanitization
  Spectre v2:            Not affected
  Srbds:                 Not affected
  Tsx async abort:       Not affected

I tried enabling them by using this command but it fails with an permission denied:

sudo su -c "echo 1 > /sys/devices/system/cpu/cpu1/online"
ash: can't create /sys/devices/system/cpu/cpu1/online: Permission denied

I researched a little bit and found out that I would need to set a Kernal parameter in order to allow "hot enabling/disabling" the CPUs. Sadly I was not able to find the location where I could do so.

Would someone know more?

EDIT: I also asked an acquaintance of mine to test this on his Raspberries. He experienced the same problems, so it is reproducible.

[mudler]

are you sure this isn't because your temperature gets high? To my experience this happens also when CPU is raising temperature and some CPUs are taken offline cause of that.

[mordax7]

I also asked an acquaintance of mine to test this on his Raspberries. Got the same result, so it is reproducible.

[mudler]

I don't have an RPI3 to test around, however with my RPI4 everything looks fine (I see all the cores onlines). Could be related to the firmware - but tricky to test on my side without dedicated hw for it. Unless you feel to try some hoops and tests by downgrading firmware and/or kernel.

[mordax7]

@mudler If you could point me in the right direction, I can try it out.

[mordax7]

Should I also open an issue?

[mudler]

yes please, let's take it from there. I'd try to downgrade firmware first - that should be rather easy as blobs can be directly replaced, then I'd try poking with kernel - I've seen regressions between firmware's more than once, so I'd start from there.