CDI spec devices are not valid input in all device input locations

[Clockwork-Muse]

/kind bug

Description

Although it is now possible to list a CDI spec during podman run, there are a few other places where devices can be specified where the CDI spec appears to not be allowed.

Steps to reproduce the issue:

1. Install podman and the nvidia-container-toolkit, and generate the relevant CDI spec.

2. Modify the containers config: sudo sed -i 's|^#devices = \[\]|devices = ["nvidia.com/gpu=gpu0"]|;' /usr/share/containers/containers.conf

3. Run a gpu-enabled container: podman run --rm docker.io/nvidia/cudagl:11.4.2-runtime-ubuntu20.04 nvidia-smi

Describe the results you received:

ERRO[0000] validating containers config: invalid device mode: nvidia.com/gpu=gpu0 

Describe the results you expected:
The container runs successfully, and produces output from nvidia-smi

Additional information you deem important (e.g. issue happens only occasionally):

I'm also attempting to use devices in a gitlab-runner via the podman socket, and something in the chain seems to be trying to stat the device line. That is, I get this error during a run:

ERROR: Job failed (system failure): prepare environment: Error response from daemon: container create: stat nvidia.com/gpu=gpu0: no such file or directory (docker.go:659:0s). Check https://docs.gitlab.com/runner/shells/index.html#shell-profile-loading for more information

... this is somewhat more unsupported, however...

Output of podman version:

Client:       Podman Engine
Version:      4.3.0-rc1
API Version:  4.3.0-rc1
Go Version:   go1.18.1
Built:        Wed Dec 31 16:00:00 1969
OS/Arch:      linux/amd64

Output of podman info:

host:
  arch: amd64
  buildahVersion: 1.28.0-dev
  cgroupControllers:
  - memory
  - pids
  cgroupManager: systemd
  cgroupVersion: v2
  conmon:
    package: conmon_2:2.1.4-0ubuntu22.04+obs13.5_amd64
    path: /usr/bin/conmon
    version: 'conmon version 2.1.4, commit: '
  cpuUtilization:
    idlePercent: 59.42
    systemPercent: 6.4
    userPercent: 34.19
  cpus: 12
  distribution:
    codename: jammy
    distribution: ubuntu
    version: "22.04"
  eventLogger: journald
  hostname: bumblebee
  idMappings:
    gidmap:
    - container_id: 0
      host_id: 1001
      size: 1
    - container_id: 1
      host_id: 165536
      size: 65536
    uidmap:
    - container_id: 0
      host_id: 1001
      size: 1
    - container_id: 1
      host_id: 165536
      size: 65536
  kernel: 5.15.0-48-generic
  linkmode: dynamic
  logDriver: journald
  memFree: 2298265600
  memTotal: 33545072640
  networkBackend: netavark
  ociRuntime:
    name: crun
    package: crun_1.6-0ubuntu22.04+obs46.2_amd64
    path: /usr/bin/crun
    version: |-
      crun version 1.6
      commit: 18cf2efbb8feb2b2f20e316520e0fd0b6c41ef4d
      spec: 1.0.0
      +SYSTEMD +SELINUX +APPARMOR +CAP +SECCOMP +EBPF +YAJL
  os: linux
  remoteSocket:
    path: /run/user/1001/podman/podman.sock
  security:
    apparmorEnabled: false
    capabilities: CAP_CHOWN,CAP_DAC_OVERRIDE,CAP_FOWNER,CAP_FSETID,CAP_KILL,CAP_NET_BIND_SERVICE,CAP_SETFCAP,CAP_SETGID,CAP_SETPCAP,CAP_SETUID,CAP_SYS_CHROOT
    rootless: true
    seccompEnabled: true
    seccompProfilePath: /usr/share/containers/seccomp.json
    selinuxEnabled: false
  serviceIsRemote: false
  slirp4netns:
    executable: /usr/bin/slirp4netns
    package: slirp4netns_1.2.0-0ubuntu22.04+obs10.4_amd64
    version: |-
      slirp4netns version 1.2.0
      commit: 656041d45cfca7a4176f6b7eed9e4fe6c11e8383
      libslirp: 4.6.1
      SLIRP_CONFIG_VERSION_MAX: 3
      libseccomp: 2.5.3
  swapFree: 753664
  swapTotal: 1023406080
  uptime: 289h 16m 2.00s (Approximately 12.04 days)
plugins:
  authorization: null
  log:
  - k8s-file
  - none
  - passthrough
  - journald
  network:
  - bridge
  - macvlan
  volume:
  - local
registries:
  search:
  - registry.fedoraproject.org
  - registry.access.redhat.com
  - docker.io
  - quay.io
store:
  configFile: /home/simhoff/.config/containers/storage.conf
  containerStore:
    number: 2
    paused: 0
    running: 1
    stopped: 1
  graphDriverName: overlay
  graphOptions: {}
  graphRoot: /home/simhoff/.local/share/containers/storage
  graphRootAllocated: 981810135040
  graphRootUsed: 486783193088
  graphStatus:
    Backing Filesystem: extfs
    Native Overlay Diff: "true"
    Supports d_type: "true"
    Using metacopy: "false"
  imageCopyTmpDir: /var/tmp
  imageStore:
    number: 291
  runRoot: /run/user/1001/containers
  volumePath: /home/simhoff/.local/share/containers/storage/volumes
version:
  APIVersion: 4.3.0-rc1
  Built: 0
  BuiltTime: Wed Dec 31 16:00:00 1969
  GitCommit: ""
  GoVersion: go1.18.1
  Os: linux
  OsArch: linux/amd64
  Version: 4.3.0-rc1

Package info (e.g. output of rpm -q podman or apt list podman or brew info podman):

Listing... Done
podman/now 4:4.3.0-rc1-0ubuntu22.04+obs60.1 amd64 [installed,local]
podman/unknown 4:4.3.0-0ubuntu22.04+obs63.1 arm64
podman/unknown 4:4.3.0-rc1-0ubuntu22.04+obs60.1 armhf
podman/unknown 4:4.3.0-rc1-0ubuntu22.04+obs60.1 s390x

Have you tested with the latest version of Podman and have you checked the Podman Troubleshooting Guide? (https://github.com/containers/podman/blob/main/troubleshooting.md)

Yes - package is from Kubic repo

Additional environment details (AWS, VirtualBox, physical, etc.):
Host OS: Ubuntu 22.04

[github-actions]

A friendly reminder that this issue had no activity for 30 days.

[rhatdan]

@giuseppe PTAL

[giuseppe]

first half of the fix: containers/common#1239

[giuseppe]

and second half: #16580