package generate import ( "context" "github.com/containers/image/v5/manifest" "github.com/containers/libpod/libpod" ann "github.com/containers/libpod/pkg/annotations" envLib "github.com/containers/libpod/pkg/env" "github.com/containers/libpod/pkg/signal" "github.com/containers/libpod/pkg/specgen" "golang.org/x/sys/unix" ) func CompleteSpec(ctx context.Context, r *libpod.Runtime, s *specgen.SpecGenerator) error { // If a rootfs is used, then there is no image data if s.ContainerStorageConfig.Rootfs != "" { return nil } newImage, err := r.ImageRuntime().NewFromLocal(s.Image) if err != nil { return err } _, mediaType, err := newImage.Manifest(ctx) if err != nil { return err } if s.HealthConfig == nil && mediaType == manifest.DockerV2Schema2MediaType { s.HealthConfig, err = newImage.GetHealthCheck(ctx) if err != nil { return err } } // Image stop signal if s.StopSignal == nil { stopSignal, err := newImage.StopSignal(ctx) if err != nil { return err } sig, err := signal.ParseSignalNameOrNumber(stopSignal) if err != nil { return err } s.StopSignal = &sig } // Image envs from the image if they don't exist // already env, err := newImage.Env(ctx) if err != nil { return err } if len(env) > 0 { envs, err := envLib.ParseSlice(env) if err != nil { return err } for k, v := range envs { if _, exists := s.Env[k]; !exists { s.Env[v] = k } } } labels, err := newImage.Labels(ctx) if err != nil { return err } // labels from the image that dont exist already for k, v := range labels { if _, exists := s.Labels[k]; !exists { s.Labels[k] = v } } // annotations // Add annotations from the image annotations, err := newImage.Annotations(ctx) if err != nil { return err } for k, v := range annotations { annotations[k] = v } // in the event this container is in a pod, and the pod has an infra container // we will want to configure it as a type "container" instead defaulting to // the behavior of a "sandbox" container // In Kata containers: // - "sandbox" is the annotation that denotes the container should use its own // VM, which is the default behavior // - "container" denotes the container should join the VM of the SandboxID // (the infra container) if len(s.Pod) > 0 { annotations[ann.SandboxID] = s.Pod annotations[ann.ContainerType] = ann.ContainerTypeContainer } // now pass in the values from client for k, v := range s.Annotations { annotations[k] = v } s.Annotations = annotations // workdir workingDir, err := newImage.WorkingDir(ctx) if err != nil { return err } if len(s.WorkDir) < 1 && len(workingDir) > 1 { s.WorkDir = workingDir } if len(s.SeccompProfilePath) < 1 { p, err := libpod.DefaultSeccompPath() if err != nil { return err } s.SeccompProfilePath = p } if len(s.User) == 0 { s.User, err = newImage.User(ctx) if err != nil { return err } // TODO This should be enabled when namespaces actually work //case usernsMode.IsKeepID(): // user = fmt.Sprintf("%d:%d", rootless.GetRootlessUID(), rootless.GetRootlessGID()) if len(s.User) == 0 { s.User = "0" } } if err := finishThrottleDevices(s); err != nil { return err } // Unless already set via the CLI, check if we need to disable process // labels or set the defaults. if len(s.SelinuxOpts) == 0 { if err := setLabelOpts(s, r, s.PidNS, s.IpcNS); err != nil { return err } } return nil } // finishThrottleDevices takes the temporary representation of the throttle // devices in the specgen and looks up the major and major minors. it then // sets the throttle devices proper in the specgen func finishThrottleDevices(s *specgen.SpecGenerator) error { if bps := s.ThrottleReadBpsDevice; len(bps) > 0 { for k, v := range bps { statT := unix.Stat_t{} if err := unix.Stat(k, &statT); err != nil { return err } v.Major = (int64(unix.Major(statT.Rdev))) v.Minor = (int64(unix.Minor(statT.Rdev))) s.ResourceLimits.BlockIO.ThrottleReadBpsDevice = append(s.ResourceLimits.BlockIO.ThrottleReadBpsDevice, v) } } if bps := s.ThrottleWriteBpsDevice; len(bps) > 0 { for k, v := range bps { statT := unix.Stat_t{} if err := unix.Stat(k, &statT); err != nil { return err } v.Major = (int64(unix.Major(statT.Rdev))) v.Minor = (int64(unix.Minor(statT.Rdev))) s.ResourceLimits.BlockIO.ThrottleWriteBpsDevice = append(s.ResourceLimits.BlockIO.ThrottleWriteBpsDevice, v) } } if iops := s.ThrottleReadIOPSDevice; len(iops) > 0 { for k, v := range iops { statT := unix.Stat_t{} if err := unix.Stat(k, &statT); err != nil { return err } v.Major = (int64(unix.Major(statT.Rdev))) v.Minor = (int64(unix.Minor(statT.Rdev))) s.ResourceLimits.BlockIO.ThrottleReadIOPSDevice = append(s.ResourceLimits.BlockIO.ThrottleReadIOPSDevice, v) } } if iops := s.ThrottleWriteBpsDevice; len(iops) > 0 { for k, v := range iops { statT := unix.Stat_t{} if err := unix.Stat(k, &statT); err != nil { return err } v.Major = (int64(unix.Major(statT.Rdev))) v.Minor = (int64(unix.Minor(statT.Rdev))) s.ResourceLimits.BlockIO.ThrottleWriteIOPSDevice = append(s.ResourceLimits.BlockIO.ThrottleWriteIOPSDevice, v) } } return nil }