cat /proc/stat shows a bunch of zeros

[thecodingwizard]

Description

When I cat /proc/stat in gvisor, I see a bunch of zeros:

cpu  0 0 0 0 0 0 0 0 0 0
cpu0 0 0 0 0 0 0 0 0 0 0
cpu1 0 0 0 0 0 0 0 0 0 0
cpu2 0 0 0 0 0 0 0 0 0 0
cpu3 0 0 0 0 0 0 0 0 0 0
cpu4 0 0 0 0 0 0 0 0 0 0
cpu5 0 0 0 0 0 0 0 0 0 0
cpu6 0 0 0 0 0 0 0 0 0 0
cpu7 0 0 0 0 0 0 0 0 0 0
intr 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
ctxt 0
btime 1706898049
processes 0
procs_running 0
procs_blocked 0
softirq 0 0 0 0 0 0 0 0 0 0 0

As such, I think tools like htop don't work (the CPU usage shown is just always zero). My goal is mostly to be able to run htop within a runsc container.

Steps to reproduce

docker run --runtime=runsc --rm -it ubuntu cat /proc/stat

You will see a bunch of zeros. Alternatively, run htop within a runsc container; the CPU usage will be zero everywhere.

runsc version

No response

docker version (if using docker)

Client: Docker Engine - Community
 Version:           24.0.7
 API version:       1.43
 Go version:        go1.20.10
 Git commit:        afdd53b
 Built:             Thu Oct 26 09:08:01 2023
 OS/Arch:           linux/amd64
 Context:           default

Server: Docker Engine - Community
 Engine:
  Version:          24.0.7
  API version:      1.43 (minimum version 1.12)
  Go version:       go1.20.10
  Git commit:       311b9ff
  Built:            Thu Oct 26 09:08:01 2023
  OS/Arch:          linux/amd64
  Experimental:     false
 containerd:
  Version:          1.6.26
  GitCommit:        3dd1e886e55dd695541fdcd67420c2888645a495
 runc:
  Version:          1.1.10
  GitCommit:        v1.1.10-0-g18a0cb0
 docker-init:
  Version:          0.19.0
  GitCommit:        de40ad0

uname

Linux ip-10-1-5-239 5.15.0-1048-aws #53~20.04.1-Ubuntu SMP Wed Oct 4 16:44:20 UTC 2023 x86_64 x86_64 x86_64 GNU/Linux

kubectl (if using Kubernetes)

No response

repo state (if built from source)

release-20240122.0-27-gac416763d

runsc debug logs (if available)

No response

[ayushr2]

This is a known deficiency:

gvisor/pkg/sentry/fsimpl/proc/tasks_files.go

Lines 187 to 195 in b09abbe

	// TODO(b/37226836): We currently export only zero CPU stats. We could
	// at least provide some aggregate stats.
	var cpu cpuStats
	fmt.Fprintf(buf, "cpu %s\n", cpu)
	k := kernel.KernelFromContext(ctx)
	for c, max := uint(0), k.ApplicationCores(); c < max; c++ {
	fmt.Fprintf(buf, "cpu%d %s\n", c, cpu)
	}

I think tools like htop don't work (the CPU usage shown is just always zero), though I'm not 100% sure that /proc/stat is the issue.

Good way to investigate this would be to run strace -f htop on host and see if it opens /proc/stat.

[thecodingwizard]

Oh, that's an excellent idea! I verified that there is a syscall to open /proc/stat when running htop. Thanks!

[thecodingwizard]

Do you know how difficult this functionality would be to implement? Would the implementation of overall CPU stats be to maintain an exitedCPUStats property in Kernel (updated every time a task exits in task_exit.go) and aggregate k.exitedCPUStats with k.TaskSet().Root.tasks.CPUStats()?

I'm not sure how to get per-core aggregate information...