diff options
36 files changed, 621 insertions, 923 deletions
diff --git a/cmd/podman/common/completion.go b/cmd/podman/common/completion.go index 5eef5f982..07dcc4e6a 100644 --- a/cmd/podman/common/completion.go +++ b/cmd/podman/common/completion.go @@ -284,7 +284,16 @@ func getNetworks(cmd *cobra.Command, toComplete string, cType completeType) ([]s return suggestions, cobra.ShellCompDirectiveNoFileComp } -func getPathCompletion(root string, toComplete string) []string { +func fdIsNotDir(f *os.File) bool { + stat, err := f.Stat() + if err != nil { + cobra.CompErrorln(err.Error()) + return true + } + return !stat.IsDir() +} + +func getPathCompletion(root string, toComplete string) ([]string, cobra.ShellCompDirective) { if toComplete == "" { toComplete = "/" } @@ -292,41 +301,61 @@ func getPathCompletion(root string, toComplete string) []string { userpath, err := securejoin.SecureJoin(root, toComplete) if err != nil { cobra.CompErrorln(err.Error()) - return nil + return nil, cobra.ShellCompDirectiveDefault } var base string f, err := os.Open(userpath) - if err != nil { + // when error or file is not dir get the parent path to stat + if err != nil || fdIsNotDir(f) { // Do not use path.Dir() since this cleans the paths which // then no longer matches the user input. userpath, base = path.Split(userpath) toComplete, _ = path.Split(toComplete) f, err = os.Open(userpath) if err != nil { - return nil + return nil, cobra.ShellCompDirectiveDefault } } - stat, err := f.Stat() - if err != nil { - cobra.CompErrorln(err.Error()) - return nil - } - if !stat.IsDir() { + + if fdIsNotDir(f) { // nothing to complete since it is no dir - return nil + return nil, cobra.ShellCompDirectiveDefault } + entries, err := f.ReadDir(-1) if err != nil { cobra.CompErrorln(err.Error()) - return nil + return nil, cobra.ShellCompDirectiveDefault + } + if len(entries) == 0 { + // path is empty dir, just add the trailing slash and no space + if !strings.HasSuffix(toComplete, "/") { + toComplete += "/" + } + return []string{toComplete}, cobra.ShellCompDirectiveDefault | cobra.ShellCompDirectiveNoSpace } completions := make([]string, 0, len(entries)) + count := 0 for _, e := range entries { if strings.HasPrefix(e.Name(), base) { - completions = append(completions, simplePathJoinUnix(toComplete, e.Name())) + suf := "" + // When the entry is an directory we add the "/" as suffix and do not want to add space + // to match normal shell completion behavior. + // Just inc counter again to fake more than one entry in this case and thus get no space. + if e.IsDir() { + suf = "/" + count++ + } + completions = append(completions, simplePathJoinUnix(toComplete, e.Name()+suf)) + count++ } } - return completions + directive := cobra.ShellCompDirectiveDefault + if count > 1 { + // when we have more than one match we do not want to add a space after the completion + directive |= cobra.ShellCompDirectiveNoSpace + } + return completions, directive } // simplePathJoinUnix joins to path components by adding a slash only if p1 doesn't end with one. @@ -605,7 +634,7 @@ func AutocompleteCreateRun(cmd *cobra.Command, args []string, toComplete string) // So this uses ShellCompDirectiveDefault to also still provide normal shell // completion in case no path matches. This is useful if someone tries to get // completion for paths that are not available in the image, e.g. /proc/... - return getPathCompletion(resp[0].Path, toComplete), cobra.ShellCompDirectiveDefault | cobra.ShellCompDirectiveNoSpace + return getPathCompletion(resp[0].Path, toComplete) } // AutocompleteRegistries - Autocomplete registries. @@ -676,7 +705,8 @@ func AutocompleteCpCommand(cmd *cobra.Command, args []string, toComplete string) if len(resp) != 1 { return nil, cobra.ShellCompDirectiveDefault } - return prefixSlice(toComplete[:i+1], getPathCompletion(resp[0].Path, toComplete[i+1:])), cobra.ShellCompDirectiveDefault | cobra.ShellCompDirectiveNoSpace + comps, directive := getPathCompletion(resp[0].Path, toComplete[i+1:]) + return prefixSlice(toComplete[:i+1], comps), directive } // Suggest containers when they match the input otherwise normal shell completion is used containers, _ := getContainers(cmd, toComplete, completeDefault) diff --git a/cmd/podman/common/create_opts.go b/cmd/podman/common/create_opts.go index c40d1ea51..ad535ff59 100644 --- a/cmd/podman/common/create_opts.go +++ b/cmd/podman/common/create_opts.go @@ -1,472 +1,9 @@ package common import ( - "fmt" - "net" - "os" - "path/filepath" - "strconv" - "strings" - - "github.com/containers/common/libnetwork/types" - "github.com/containers/common/pkg/cgroups" - "github.com/containers/common/pkg/config" "github.com/containers/podman/v4/cmd/podman/registry" - "github.com/containers/podman/v4/libpod/define" - "github.com/containers/podman/v4/pkg/api/handlers" - "github.com/containers/podman/v4/pkg/domain/entities" - "github.com/containers/podman/v4/pkg/rootless" - "github.com/containers/podman/v4/pkg/specgen" - "github.com/docker/docker/api/types/mount" - "github.com/pkg/errors" ) -func stringMaptoArray(m map[string]string) []string { - a := make([]string, 0, len(m)) - for k, v := range m { - a = append(a, fmt.Sprintf("%s=%s", k, v)) - } - return a -} - -// ContainerCreateToContainerCLIOpts converts a compat input struct to cliopts so it can be converted to -// a specgen spec. -func ContainerCreateToContainerCLIOpts(cc handlers.CreateContainerConfig, rtc *config.Config) (*entities.ContainerCreateOptions, []string, error) { - var ( - capAdd []string - cappDrop []string - entrypoint *string - init bool - specPorts []types.PortMapping - ) - - if cc.HostConfig.Init != nil { - init = *cc.HostConfig.Init - } - - // Iterate devices and convert back to string - devices := make([]string, 0, len(cc.HostConfig.Devices)) - for _, dev := range cc.HostConfig.Devices { - devices = append(devices, fmt.Sprintf("%s:%s:%s", dev.PathOnHost, dev.PathInContainer, dev.CgroupPermissions)) - } - - // iterate blkreaddevicebps - readBps := make([]string, 0, len(cc.HostConfig.BlkioDeviceReadBps)) - for _, dev := range cc.HostConfig.BlkioDeviceReadBps { - readBps = append(readBps, dev.String()) - } - - // iterate blkreaddeviceiops - readIops := make([]string, 0, len(cc.HostConfig.BlkioDeviceReadIOps)) - for _, dev := range cc.HostConfig.BlkioDeviceReadIOps { - readIops = append(readIops, dev.String()) - } - - // iterate blkwritedevicebps - writeBps := make([]string, 0, len(cc.HostConfig.BlkioDeviceWriteBps)) - for _, dev := range cc.HostConfig.BlkioDeviceWriteBps { - writeBps = append(writeBps, dev.String()) - } - - // iterate blkwritedeviceiops - writeIops := make([]string, 0, len(cc.HostConfig.BlkioDeviceWriteIOps)) - for _, dev := range cc.HostConfig.BlkioDeviceWriteIOps { - writeIops = append(writeIops, dev.String()) - } - - // entrypoint - // can be a string or slice. if it is a slice, we need to - // marshall it to json; otherwise it should just be the string - // value - if len(cc.Config.Entrypoint) > 0 { - entrypoint = &cc.Config.Entrypoint[0] - if len(cc.Config.Entrypoint) > 1 { - b, err := json.Marshal(cc.Config.Entrypoint) - if err != nil { - return nil, nil, err - } - var jsonString = string(b) - entrypoint = &jsonString - } - } - - // expose ports - expose := make([]string, 0, len(cc.Config.ExposedPorts)) - for p := range cc.Config.ExposedPorts { - expose = append(expose, fmt.Sprintf("%s/%s", p.Port(), p.Proto())) - } - - // mounts type=tmpfs/bind,source=...,target=...=,opt=val - volSources := make(map[string]bool) - volDestinations := make(map[string]bool) - mounts := make([]string, 0, len(cc.HostConfig.Mounts)) - var builder strings.Builder - for _, m := range cc.HostConfig.Mounts { - addField(&builder, "type", string(m.Type)) - addField(&builder, "source", m.Source) - addField(&builder, "target", m.Target) - - // Store source/dest so we don't add duplicates if a volume is - // also mentioned in cc.Volumes. - // Which Docker Compose v2.0 does, for unclear reasons... - volSources[m.Source] = true - volDestinations[m.Target] = true - - if m.ReadOnly { - addField(&builder, "ro", "true") - } - addField(&builder, "consistency", string(m.Consistency)) - // Map any specialized mount options that intersect between *Options and cli options - switch m.Type { - case mount.TypeBind: - if m.BindOptions != nil { - addField(&builder, "bind-propagation", string(m.BindOptions.Propagation)) - addField(&builder, "bind-nonrecursive", strconv.FormatBool(m.BindOptions.NonRecursive)) - } - case mount.TypeTmpfs: - if m.TmpfsOptions != nil { - addField(&builder, "tmpfs-size", strconv.FormatInt(m.TmpfsOptions.SizeBytes, 10)) - addField(&builder, "tmpfs-mode", strconv.FormatUint(uint64(m.TmpfsOptions.Mode), 8)) - } - case mount.TypeVolume: - // All current VolumeOpts are handled above - // See vendor/github.com/containers/common/pkg/parse/parse.go:ValidateVolumeOpts() - } - mounts = append(mounts, builder.String()) - builder.Reset() - } - - // dns - dns := make([]net.IP, 0, len(cc.HostConfig.DNS)) - for _, d := range cc.HostConfig.DNS { - dns = append(dns, net.ParseIP(d)) - } - - // publish - for port, pbs := range cc.HostConfig.PortBindings { - for _, pb := range pbs { - var hostport int - var err error - if pb.HostPort != "" { - hostport, err = strconv.Atoi(pb.HostPort) - } - if err != nil { - return nil, nil, err - } - tmpPort := types.PortMapping{ - HostIP: pb.HostIP, - ContainerPort: uint16(port.Int()), - HostPort: uint16(hostport), - Range: 0, - Protocol: port.Proto(), - } - specPorts = append(specPorts, tmpPort) - } - } - - // special case for NetworkMode, the podman default is slirp4netns for - // rootless but for better docker compat we want bridge. - netmode := string(cc.HostConfig.NetworkMode) - if netmode == "" || netmode == "default" { - netmode = "bridge" - } - nsmode, networks, netOpts, err := specgen.ParseNetworkFlag([]string{netmode}) - if err != nil { - return nil, nil, err - } - - // network - // Note: we cannot emulate compat exactly here. we only allow specifics of networks to be - // defined when there is only one network. - netInfo := entities.NetOptions{ - AddHosts: cc.HostConfig.ExtraHosts, - DNSOptions: cc.HostConfig.DNSOptions, - DNSSearch: cc.HostConfig.DNSSearch, - DNSServers: dns, - Network: nsmode, - PublishPorts: specPorts, - NetworkOptions: netOpts, - NoHosts: rtc.Containers.NoHosts, - } - - // network names - switch { - case len(cc.NetworkingConfig.EndpointsConfig) > 0: - endpointsConfig := cc.NetworkingConfig.EndpointsConfig - networks := make(map[string]types.PerNetworkOptions, len(endpointsConfig)) - for netName, endpoint := range endpointsConfig { - netOpts := types.PerNetworkOptions{} - if endpoint != nil { - netOpts.Aliases = endpoint.Aliases - - // if IP address is provided - if len(endpoint.IPAddress) > 0 { - staticIP := net.ParseIP(endpoint.IPAddress) - if staticIP == nil { - return nil, nil, errors.Errorf("failed to parse the ip address %q", endpoint.IPAddress) - } - netOpts.StaticIPs = append(netOpts.StaticIPs, staticIP) - } - - if endpoint.IPAMConfig != nil { - // if IPAMConfig.IPv4Address is provided - if len(endpoint.IPAMConfig.IPv4Address) > 0 { - staticIP := net.ParseIP(endpoint.IPAMConfig.IPv4Address) - if staticIP == nil { - return nil, nil, errors.Errorf("failed to parse the ipv4 address %q", endpoint.IPAMConfig.IPv4Address) - } - netOpts.StaticIPs = append(netOpts.StaticIPs, staticIP) - } - // if IPAMConfig.IPv6Address is provided - if len(endpoint.IPAMConfig.IPv6Address) > 0 { - staticIP := net.ParseIP(endpoint.IPAMConfig.IPv6Address) - if staticIP == nil { - return nil, nil, errors.Errorf("failed to parse the ipv6 address %q", endpoint.IPAMConfig.IPv6Address) - } - netOpts.StaticIPs = append(netOpts.StaticIPs, staticIP) - } - } - // If MAC address is provided - if len(endpoint.MacAddress) > 0 { - staticMac, err := net.ParseMAC(endpoint.MacAddress) - if err != nil { - return nil, nil, errors.Errorf("failed to parse the mac address %q", endpoint.MacAddress) - } - netOpts.StaticMAC = types.HardwareAddr(staticMac) - } - } - - networks[netName] = netOpts - } - - netInfo.Networks = networks - case len(cc.HostConfig.NetworkMode) > 0: - netInfo.Networks = networks - } - - parsedTmp := make([]string, 0, len(cc.HostConfig.Tmpfs)) - for path, options := range cc.HostConfig.Tmpfs { - finalString := path - if options != "" { - finalString += ":" + options - } - parsedTmp = append(parsedTmp, finalString) - } - - // Note: several options here are marked as "don't need". this is based - // on speculation by Matt and I. We think that these come into play later - // like with start. We believe this is just a difference in podman/compat - cliOpts := entities.ContainerCreateOptions{ - // Attach: nil, // don't need? - Authfile: "", - CapAdd: append(capAdd, cc.HostConfig.CapAdd...), - CapDrop: append(cappDrop, cc.HostConfig.CapDrop...), - CgroupParent: cc.HostConfig.CgroupParent, - CIDFile: cc.HostConfig.ContainerIDFile, - CPUPeriod: uint64(cc.HostConfig.CPUPeriod), - CPUQuota: cc.HostConfig.CPUQuota, - CPURTPeriod: uint64(cc.HostConfig.CPURealtimePeriod), - CPURTRuntime: cc.HostConfig.CPURealtimeRuntime, - CPUShares: uint64(cc.HostConfig.CPUShares), - // CPUS: 0, // don't need? - CPUSetCPUs: cc.HostConfig.CpusetCpus, - CPUSetMems: cc.HostConfig.CpusetMems, - // Detach: false, // don't need - // DetachKeys: "", // don't need - Devices: devices, - DeviceCgroupRule: nil, - DeviceReadBPs: readBps, - DeviceReadIOPs: readIops, - DeviceWriteBPs: writeBps, - DeviceWriteIOPs: writeIops, - Entrypoint: entrypoint, - Env: cc.Config.Env, - Expose: expose, - GroupAdd: cc.HostConfig.GroupAdd, - Hostname: cc.Config.Hostname, - ImageVolume: "bind", - Init: init, - Interactive: cc.Config.OpenStdin, - IPC: string(cc.HostConfig.IpcMode), - Label: stringMaptoArray(cc.Config.Labels), - LogDriver: cc.HostConfig.LogConfig.Type, - LogOptions: stringMaptoArray(cc.HostConfig.LogConfig.Config), - Name: cc.Name, - OOMScoreAdj: &cc.HostConfig.OomScoreAdj, - Arch: "", - OS: "", - Variant: "", - PID: string(cc.HostConfig.PidMode), - PIDsLimit: cc.HostConfig.PidsLimit, - Privileged: cc.HostConfig.Privileged, - PublishAll: cc.HostConfig.PublishAllPorts, - Quiet: false, - ReadOnly: cc.HostConfig.ReadonlyRootfs, - ReadOnlyTmpFS: true, // podman default - Rm: cc.HostConfig.AutoRemove, - SecurityOpt: cc.HostConfig.SecurityOpt, - StopSignal: cc.Config.StopSignal, - StorageOpts: stringMaptoArray(cc.HostConfig.StorageOpt), - Sysctl: stringMaptoArray(cc.HostConfig.Sysctls), - Systemd: "true", // podman default - TmpFS: parsedTmp, - TTY: cc.Config.Tty, - UnsetEnv: cc.UnsetEnv, - UnsetEnvAll: cc.UnsetEnvAll, - User: cc.Config.User, - UserNS: string(cc.HostConfig.UsernsMode), - UTS: string(cc.HostConfig.UTSMode), - Mount: mounts, - VolumesFrom: cc.HostConfig.VolumesFrom, - Workdir: cc.Config.WorkingDir, - Net: &netInfo, - HealthInterval: define.DefaultHealthCheckInterval, - HealthRetries: define.DefaultHealthCheckRetries, - HealthTimeout: define.DefaultHealthCheckTimeout, - HealthStartPeriod: define.DefaultHealthCheckStartPeriod, - } - if !rootless.IsRootless() { - var ulimits []string - if len(cc.HostConfig.Ulimits) > 0 { - for _, ul := range cc.HostConfig.Ulimits { - ulimits = append(ulimits, ul.String()) - } - cliOpts.Ulimit = ulimits - } - } - if cc.HostConfig.Resources.NanoCPUs > 0 { - if cliOpts.CPUPeriod != 0 || cliOpts.CPUQuota != 0 { - return nil, nil, errors.Errorf("NanoCpus conflicts with CpuPeriod and CpuQuota") - } - cliOpts.CPUPeriod = 100000 - cliOpts.CPUQuota = cc.HostConfig.Resources.NanoCPUs / 10000 - } - - // volumes - for _, vol := range cc.HostConfig.Binds { - cliOpts.Volume = append(cliOpts.Volume, vol) - // Extract the destination so we don't add duplicate mounts in - // the volumes phase. - splitVol := specgen.SplitVolumeString(vol) - switch len(splitVol) { - case 1: - volDestinations[vol] = true - default: - volSources[splitVol[0]] = true - volDestinations[splitVol[1]] = true - } - } - // Anonymous volumes are added differently from other volumes, in their - // own special field, for reasons known only to Docker. Still use the - // format of `-v` so we can just append them in there. - // Unfortunately, these may be duplicates of existing mounts in Binds. - // So... We need to catch that. - // This also handles volumes duplicated between cc.HostConfig.Mounts and - // cc.Volumes, as seen in compose v2.0. - for vol := range cc.Volumes { - if _, ok := volDestinations[filepath.Clean(vol)]; ok { - continue - } - cliOpts.Volume = append(cliOpts.Volume, vol) - } - // Make mount points for compat volumes - for vol := range volSources { - // This might be a named volume. - // Assume it is if it's not an absolute path. - if !filepath.IsAbs(vol) { - continue - } - // If volume already exists, there is nothing to do - if _, err := os.Stat(vol); err == nil { - continue - } - if err := os.MkdirAll(vol, 0755); err != nil { - if !os.IsExist(err) { - return nil, nil, errors.Wrapf(err, "error making volume mountpoint for volume %s", vol) - } - } - } - if len(cc.HostConfig.BlkioWeightDevice) > 0 { - devices := make([]string, 0, len(cc.HostConfig.BlkioWeightDevice)) - for _, d := range cc.HostConfig.BlkioWeightDevice { - devices = append(devices, d.String()) - } - cliOpts.BlkIOWeightDevice = devices - } - if cc.HostConfig.BlkioWeight > 0 { - cliOpts.BlkIOWeight = strconv.Itoa(int(cc.HostConfig.BlkioWeight)) - } - - if cc.HostConfig.Memory > 0 { - cliOpts.Memory = strconv.Itoa(int(cc.HostConfig.Memory)) - } - - if cc.HostConfig.MemoryReservation > 0 { - cliOpts.MemoryReservation = strconv.Itoa(int(cc.HostConfig.MemoryReservation)) - } - - cgroupsv2, err := cgroups.IsCgroup2UnifiedMode() - if err != nil { - return nil, nil, err - } - if cc.HostConfig.MemorySwap > 0 && (!rootless.IsRootless() || (rootless.IsRootless() && cgroupsv2)) { - cliOpts.MemorySwap = strconv.Itoa(int(cc.HostConfig.MemorySwap)) - } - - if cc.Config.StopTimeout != nil { - cliOpts.StopTimeout = uint(*cc.Config.StopTimeout) - } - - if cc.HostConfig.ShmSize > 0 { - cliOpts.ShmSize = strconv.Itoa(int(cc.HostConfig.ShmSize)) - } - - if len(cc.HostConfig.RestartPolicy.Name) > 0 { - policy := cc.HostConfig.RestartPolicy.Name - // only add restart count on failure - if cc.HostConfig.RestartPolicy.IsOnFailure() { - policy += fmt.Sprintf(":%d", cc.HostConfig.RestartPolicy.MaximumRetryCount) - } - cliOpts.Restart = policy - } - - if cc.HostConfig.MemorySwappiness != nil && (!rootless.IsRootless() || rootless.IsRootless() && cgroupsv2 && rtc.Engine.CgroupManager == "systemd") { - cliOpts.MemorySwappiness = *cc.HostConfig.MemorySwappiness - } else { - cliOpts.MemorySwappiness = -1 - } - if cc.HostConfig.OomKillDisable != nil { - cliOpts.OOMKillDisable = *cc.HostConfig.OomKillDisable - } - if cc.Config.Healthcheck != nil { - finCmd := "" - for _, str := range cc.Config.Healthcheck.Test { - finCmd = finCmd + str + " " - } - if len(finCmd) > 1 { - finCmd = finCmd[:len(finCmd)-1] - } - cliOpts.HealthCmd = finCmd - if cc.Config.Healthcheck.Interval > 0 { - cliOpts.HealthInterval = cc.Config.Healthcheck.Interval.String() - } - if cc.Config.Healthcheck.Retries > 0 { - cliOpts.HealthRetries = uint(cc.Config.Healthcheck.Retries) - } - if cc.Config.Healthcheck.StartPeriod > 0 { - cliOpts.HealthStartPeriod = cc.Config.Healthcheck.StartPeriod.String() - } - if cc.Config.Healthcheck.Timeout > 0 { - cliOpts.HealthTimeout = cc.Config.Healthcheck.Timeout.String() - } - } - - // specgen assumes the image name is arg[0] - cmd := []string{cc.Config.Image} - cmd = append(cmd, cc.Config.Cmd...) - return &cliOpts, cmd, nil -} - func ulimits() []string { if !registry.IsRemote() { return containerConfig.Ulimits() @@ -536,17 +73,3 @@ func LogDriver() string { } return "" } - -// addField is a helper function to populate mount options -func addField(b *strings.Builder, name string, value string) { - if value == "" { - return - } - - if b.Len() > 0 { - b.WriteRune(',') - } - b.WriteString(name) - b.WriteRune('=') - b.WriteString(value) -} diff --git a/docs/source/markdown/podman-create.1.md b/docs/source/markdown/podman-create.1.md index 913183869..d5e96168d 100644 --- a/docs/source/markdown/podman-create.1.md +++ b/docs/source/markdown/podman-create.1.md @@ -1261,9 +1261,9 @@ Podman allocates unique ranges of UIDs and GIDs from the `containers` subordinat **host**: run in the user namespace of the caller. The processes running in the container will have the same privileges on the host as any other process launched by the calling user (default). -**keep-id**: creates a user namespace where the current rootless user's UID:GID are mapped to the same values in the container. This option is ignored for containers created by the root user. +**keep-id**: creates a user namespace where the current rootless user's UID:GID are mapped to the same values in the container. This option is not allowed for containers created by the root user. -**nomap**: creates a user namespace where the current rootless user's UID:GID are not mapped into the container. This option is ignored for containers created by the root user. +**nomap**: creates a user namespace where the current rootless user's UID:GID are not mapped into the container. This option is not allowed for containers created by the root user. **ns:**_namespace_: run the container in the given existing user namespace. diff --git a/docs/source/markdown/podman-play-kube.1.md b/docs/source/markdown/podman-play-kube.1.md index 08bb2a5bc..1c7fc99a2 100644 --- a/docs/source/markdown/podman-play-kube.1.md +++ b/docs/source/markdown/podman-play-kube.1.md @@ -276,9 +276,9 @@ Podman allocates unique ranges of UIDs and GIDs from the `containers` subordinat **host**: create a new namespace for the container. -**keep-id**: creates a user namespace where the current rootless user's UID:GID are mapped to the same values in the container. This option is ignored for containers created by the root user. +**keep-id**: creates a user namespace where the current rootless user's UID:GID are mapped to the same values in the container. This option is not allowed for containers created by the root user. -**nomap**: creates a user namespace where the current rootless user's UID:GID are not mapped into the container. This option is ignored for containers created by the root user. +**nomap**: creates a user namespace where the current rootless user's UID:GID are not mapped into the container. This option is not allowed for containers created by the root user. **ns:**_namespace_: run the pod in the given existing user namespace. diff --git a/docs/source/markdown/podman-pod-create.1.md b/docs/source/markdown/podman-pod-create.1.md index 1a98528bb..d5b8232cf 100644 --- a/docs/source/markdown/podman-pod-create.1.md +++ b/docs/source/markdown/podman-pod-create.1.md @@ -360,9 +360,9 @@ Valid _mode_ values are: - *host*: run in the user namespace of the caller. The processes running in the container will have the same privileges on the host as any other process launched by the calling user (default). - - *keep-id*: creates a user namespace where the current rootless user's UID:GID are mapped to the same values in the container. This option is ignored for containers created by the root user. + - *keep-id*: creates a user namespace where the current rootless user's UID:GID are mapped to the same values in the container. This option is not allowed for containers created by the root user. - - *nomap*: creates a user namespace where the current rootless user's UID:GID are not mapped into the container. This option is ignored for containers created by the root user. + - *nomap*: creates a user namespace where the current rootless user's UID:GID are not mapped into the container. This option is not allowed for containers created by the root user. #### **--volume**, **-v**[=*[[SOURCE-VOLUME|HOST-DIR:]CONTAINER-DIR[:OPTIONS]]*] diff --git a/docs/source/markdown/podman-run.1.md b/docs/source/markdown/podman-run.1.md index b9d87b5bd..6acbec7d5 100644 --- a/docs/source/markdown/podman-run.1.md +++ b/docs/source/markdown/podman-run.1.md @@ -1329,9 +1329,9 @@ The rootless option `--userns=keep-id` uses all the subuids and subgids of the u **host**: run in the user namespace of the caller. The processes running in the container will have the same privileges on the host as any other process launched by the calling user (default). -**keep-id**: creates a user namespace where the current rootless user's UID:GID are mapped to the same values in the container. This option is ignored for containers created by the root user. +**keep-id**: creates a user namespace where the current rootless user's UID:GID are mapped to the same values in the container. This option is not allowed for containers created by the root user. -**nomap**: creates a user namespace where the current rootless user's UID:GID are not mapped into the container. This option is ignored for containers created by the root user. +**nomap**: creates a user namespace where the current rootless user's UID:GID are not mapped into the container. This option is not allowed for containers created by the root user. **ns:**_namespace_: run the container in the given existing user namespace. @@ -47,7 +47,7 @@ require ( github.com/onsi/gomega v1.19.0 github.com/opencontainers/go-digest v1.0.0 github.com/opencontainers/image-spec v1.0.3-0.20220114050600-8b9d41f48198 - github.com/opencontainers/runc v1.1.2 + github.com/opencontainers/runc v1.1.3 github.com/opencontainers/runtime-spec v1.0.3-0.20211214071223-8958f93039ab github.com/opencontainers/runtime-tools v0.9.1-0.20220110225228-7e2d60f1e41f github.com/opencontainers/selinux v1.10.1 @@ -1054,8 +1054,9 @@ github.com/opencontainers/runc v1.0.2/go.mod h1:aTaHFFwQXuA71CiyxOdFFIorAoemI04s github.com/opencontainers/runc v1.0.3/go.mod h1:aTaHFFwQXuA71CiyxOdFFIorAoemI04suvGRQFzWTD0= github.com/opencontainers/runc v1.1.0/go.mod h1:Tj1hFw6eFWp/o33uxGf5yF2BX5yz2Z6iptFpuvbbKqc= github.com/opencontainers/runc v1.1.1/go.mod h1:Tj1hFw6eFWp/o33uxGf5yF2BX5yz2Z6iptFpuvbbKqc= -github.com/opencontainers/runc v1.1.2 h1:2VSZwLx5k/BfsBxMMipG/LYUnmqOD/BPkIVgQUcTlLw= github.com/opencontainers/runc v1.1.2/go.mod h1:Tj1hFw6eFWp/o33uxGf5yF2BX5yz2Z6iptFpuvbbKqc= +github.com/opencontainers/runc v1.1.3 h1:vIXrkId+0/J2Ymu2m7VjGvbSlAId9XNRPhn2p4b+d8w= +github.com/opencontainers/runc v1.1.3/go.mod h1:1J5XiS+vdZ3wCyZybsuxXZWGrgSr8fFJHLXuG2PsnNg= github.com/opencontainers/runtime-spec v0.1.2-0.20190507144316-5b71a03e2700/go.mod h1:jwyrGlmzljRJv/Fgzds9SsS/C5hL+LL3ko9hs6T5lQ0= github.com/opencontainers/runtime-spec v1.0.1/go.mod h1:jwyrGlmzljRJv/Fgzds9SsS/C5hL+LL3ko9hs6T5lQ0= github.com/opencontainers/runtime-spec v1.0.2-0.20190207185410-29686dbc5559/go.mod h1:jwyrGlmzljRJv/Fgzds9SsS/C5hL+LL3ko9hs6T5lQ0= @@ -1187,8 +1188,9 @@ github.com/sean-/seed v0.0.0-20170313163322-e2103e2c3529/go.mod h1:DxrIzT+xaE7yg github.com/sebdah/goldie/v2 v2.5.3 h1:9ES/mNN+HNUbNWpVAlrzuZ7jE+Nrczbj8uFRjM7624Y= github.com/sebdah/goldie/v2 v2.5.3/go.mod h1:oZ9fp0+se1eapSRjfYbsV/0Hqhbuu3bJVvKI/NNtssI= github.com/seccomp/libseccomp-golang v0.9.1/go.mod h1:GbW5+tmTXfcxTToHLXlScSlAvWlF4P2Ca7zGrPiEpWo= -github.com/seccomp/libseccomp-golang v0.9.2-0.20210429002308-3879420cc921 h1:58EBmR2dMNL2n/FnbQewK3D14nXr0V9CObDSvMJLq+Y= github.com/seccomp/libseccomp-golang v0.9.2-0.20210429002308-3879420cc921/go.mod h1:JA8cRccbGaA1s33RQf7Y1+q9gHmZX1yB/z9WDN1C6fg= +github.com/seccomp/libseccomp-golang v0.9.2-0.20220502022130-f33da4d89646 h1:RpforrEYXWkmGwJHIGnLZ3tTWStkjVVstwzNGqxX2Ds= +github.com/seccomp/libseccomp-golang v0.9.2-0.20220502022130-f33da4d89646/go.mod h1:JA8cRccbGaA1s33RQf7Y1+q9gHmZX1yB/z9WDN1C6fg= github.com/securego/gosec/v2 v2.9.1/go.mod h1:oDcDLcatOJxkCGaCaq8lua1jTnYf6Sou4wdiJ1n4iHc= github.com/sergi/go-diff v1.0.0/go.mod h1:0CfEIISq7TuYL3j771MWULgwwjU+GofnZX9QAmXWZgo= github.com/sergi/go-diff v1.1.0/go.mod h1:STckp+ISIX8hZLjrqAeVduY0gWCT9IjLuqbuNXdaHfM= diff --git a/libpod/oci_conmon_linux.go b/libpod/oci_conmon_linux.go index 6aa7ce6dc..0c1ee61d3 100644 --- a/libpod/oci_conmon_linux.go +++ b/libpod/oci_conmon_linux.go @@ -411,8 +411,8 @@ func (r *ConmonOCIRuntime) KillContainer(ctr *Container, signal uint, all bool) if err2 := r.UpdateContainerStatus(ctr); err2 != nil { logrus.Infof("Error updating status for container %s: %v", ctr.ID(), err2) } - if ctr.state.State == define.ContainerStateExited { - return nil + if ctr.ensureState(define.ContainerStateStopped, define.ContainerStateExited) { + return define.ErrCtrStateInvalid } return errors.Wrapf(err, "error sending signal to container %s", ctr.ID()) } diff --git a/libpod/pod.go b/libpod/pod.go index 3c8dc43d4..108317637 100644 --- a/libpod/pod.go +++ b/libpod/pod.go @@ -178,8 +178,8 @@ func (p *Pod) NetworkMode() string { return infra.NetworkMode() } -// PidMode returns the PID mode given by the user ex: pod, private... -func (p *Pod) PidMode() string { +// Namespace Mode returns the given NS mode provided by the user ex: host, private... +func (p *Pod) NamespaceMode(kind specs.LinuxNamespaceType) string { infra, err := p.runtime.GetContainer(p.state.InfraContainerID) if err != nil { return "" @@ -187,28 +187,7 @@ func (p *Pod) PidMode() string { ctrSpec := infra.config.Spec if ctrSpec != nil && ctrSpec.Linux != nil { for _, ns := range ctrSpec.Linux.Namespaces { - if ns.Type == specs.PIDNamespace { - if ns.Path != "" { - return fmt.Sprintf("ns:%s", ns.Path) - } - return "private" - } - } - return "host" - } - return "" -} - -// PidMode returns the PID mode given by the user ex: pod, private... -func (p *Pod) UserNSMode() string { - infra, err := p.infraContainer() - if err != nil { - return "" - } - ctrSpec := infra.config.Spec - if ctrSpec != nil && ctrSpec.Linux != nil { - for _, ns := range ctrSpec.Linux.Namespaces { - if ns.Type == specs.UserNamespace { + if ns.Type == kind { if ns.Path != "" { return fmt.Sprintf("ns:%s", ns.Path) } diff --git a/libpod/pod_api.go b/libpod/pod_api.go index 1c1e15984..fefe0e329 100644 --- a/libpod/pod_api.go +++ b/libpod/pod_api.go @@ -9,6 +9,7 @@ import ( "github.com/containers/podman/v4/libpod/events" "github.com/containers/podman/v4/pkg/parallel" "github.com/containers/podman/v4/pkg/rootless" + "github.com/opencontainers/runtime-spec/specs-go" "github.com/pkg/errors" "github.com/sirupsen/logrus" ) @@ -673,8 +674,8 @@ func (p *Pod) Inspect() (*define.InspectPodData, error) { infraConfig.CPUPeriod = p.CPUPeriod() infraConfig.CPUQuota = p.CPUQuota() infraConfig.CPUSetCPUs = p.ResourceLim().CPU.Cpus - infraConfig.PidNS = p.PidMode() - infraConfig.UserNS = p.UserNSMode() + infraConfig.PidNS = p.NamespaceMode(specs.PIDNamespace) + infraConfig.UserNS = p.NamespaceMode(specs.UserNamespace) namedVolumes, mounts := infra.SortUserVolumes(infra.config.Spec) inspectMounts, err = infra.GetMounts(namedVolumes, infra.config.ImageVolumes, mounts) infraSecurity = infra.GetSecurityOptions() diff --git a/pkg/api/handlers/compat/containers_create.go b/pkg/api/handlers/compat/containers_create.go index b9b7f6708..67ec52047 100644 --- a/pkg/api/handlers/compat/containers_create.go +++ b/pkg/api/handlers/compat/containers_create.go @@ -261,8 +261,13 @@ func cliOpts(cc handlers.CreateContainerConfig, rtc *config.Config) (*entities.C } } - // netMode - nsmode, networks, netOpts, err := specgen.ParseNetworkFlag([]string{string(cc.HostConfig.NetworkMode)}) + // special case for NetworkMode, the podman default is slirp4netns for + // rootless but for better docker compat we want bridge. + netmode := string(cc.HostConfig.NetworkMode) + if netmode == "" || netmode == "default" { + netmode = "bridge" + } + nsmode, networks, netOpts, err := specgen.ParseNetworkFlag([]string{netmode}) if err != nil { return nil, nil, err } @@ -278,6 +283,7 @@ func cliOpts(cc handlers.CreateContainerConfig, rtc *config.Config) (*entities.C Network: nsmode, PublishPorts: specPorts, NetworkOptions: netOpts, + NoHosts: rtc.Containers.NoHosts, } // network names @@ -438,7 +444,7 @@ func cliOpts(cc handlers.CreateContainerConfig, rtc *config.Config) (*entities.C cliOpts.Volume = append(cliOpts.Volume, vol) // Extract the destination so we don't add duplicate mounts in // the volumes phase. - splitVol := strings.SplitN(vol, ":", 3) + splitVol := specgen.SplitVolumeString(vol) switch len(splitVol) { case 1: volDestinations[vol] = true diff --git a/pkg/api/handlers/libpod/manifests.go b/pkg/api/handlers/libpod/manifests.go index 65b9d6cb5..d9ed1c265 100644 --- a/pkg/api/handlers/libpod/manifests.go +++ b/pkg/api/handlers/libpod/manifests.go @@ -163,7 +163,6 @@ func ManifestAddV3(w http.ResponseWriter, r *http.Request) { // Wrapper to support 3.x with 4.x libpod query := struct { entities.ManifestAddOptions - Images []string TLSVerify bool `schema:"tlsVerify"` }{} if err := json.NewDecoder(r.Body).Decode(&query); err != nil { diff --git a/pkg/domain/infra/abi/containers.go b/pkg/domain/infra/abi/containers.go index d2fafccb1..8bd84a310 100644 --- a/pkg/domain/infra/abi/containers.go +++ b/pkg/domain/infra/abi/containers.go @@ -616,6 +616,7 @@ func (ic *ContainerEngine) ContainerRestore(ctx context.Context, namesOrIds []st ImportPrevious: options.ImportPrevious, Pod: options.Pod, PrintStats: options.PrintStats, + FileLocks: options.FileLocks, } filterFuncs := []libpod.ContainerFilter{ diff --git a/pkg/specgen/generate/namespaces.go b/pkg/specgen/generate/namespaces.go index 4735111c8..4224d16ce 100644 --- a/pkg/specgen/generate/namespaces.go +++ b/pkg/specgen/generate/namespaces.go @@ -19,6 +19,8 @@ import ( "github.com/sirupsen/logrus" ) +const host = "host" + // Get the default namespace mode for any given namespace type. func GetDefaultNamespaceMode(nsType string, cfg *config.Config, pod *libpod.Pod) (specgen.Namespace, error) { // The default for most is private @@ -33,19 +35,38 @@ func GetDefaultNamespaceMode(nsType string, cfg *config.Config, pod *libpod.Pod) podMode := false switch { case nsType == "pid" && pod.SharesPID(): + if pod.NamespaceMode(spec.PIDNamespace) == host { + toReturn.NSMode = specgen.Host + return toReturn, nil + } podMode = true case nsType == "ipc" && pod.SharesIPC(): + if pod.NamespaceMode(spec.IPCNamespace) == host { + toReturn.NSMode = specgen.Host + return toReturn, nil + } podMode = true case nsType == "uts" && pod.SharesUTS(): + if pod.NamespaceMode(spec.UTSNamespace) == host { + toReturn.NSMode = specgen.Host + return toReturn, nil + } podMode = true case nsType == "user" && pod.SharesUser(): + // user does not need a special check for host, this is already validated on pod creation + // if --userns=host then pod.SharesUser == false podMode = true case nsType == "net" && pod.SharesNet(): + if pod.NetworkMode() == host { + toReturn.NSMode = specgen.Host + return toReturn, nil + } podMode = true - case nsType == "net" && pod.NetworkMode() == "host": - toReturn.NSMode = specgen.Host - return toReturn, nil case nsType == "cgroup" && pod.SharesCgroup(): + if pod.NamespaceMode(spec.CgroupNamespace) == host { + toReturn.NSMode = specgen.Host + return toReturn, nil + } podMode = true } if podMode { @@ -491,10 +512,7 @@ func GetNamespaceOptions(ns []string, netnsIsHost bool) ([]libpod.PodCreateOptio case "cgroup": options = append(options, libpod.WithPodCgroup()) case "net": - // share the netns setting with other containers in the pod only when it is not set to host - if !netnsIsHost { - options = append(options, libpod.WithPodNet()) - } + options = append(options, libpod.WithPodNet()) case "mnt": return erroredOptions, errors.Errorf("Mount sharing functionality not supported on pod level") case "pid": diff --git a/test/apiv2/10-images.at b/test/apiv2/10-images.at index 13aaea317..f03b95786 100644 --- a/test/apiv2/10-images.at +++ b/test/apiv2/10-images.at @@ -53,7 +53,7 @@ t POST "images/create?fromImage=alpine" 200 .error~null .status~".*Download comp t POST "images/create?fromImage=alpine&tag=latest" 200 # 10977 - handle platform parameter correctly -t POST "images/create?fromImage=testimage:20210610&platform=linux/arm64" 200 +t POST "images/create?fromImage=quay.io/libpod/testimage:20210610&platform=linux/arm64" 200 t GET "images/testimage:20210610/json" 200 \ .Architecture=arm64 diff --git a/test/apiv2/12-imagesMore.at b/test/apiv2/12-imagesMore.at index 67b4f1c79..57d5e114d 100644 --- a/test/apiv2/12-imagesMore.at +++ b/test/apiv2/12-imagesMore.at @@ -6,6 +6,8 @@ red='\e[31m' nc='\e[0m' +start_registry + podman pull -q $IMAGE t GET libpod/images/json 200 \ @@ -20,48 +22,38 @@ t GET libpod/images/$IMAGE/tree 200 \ t POST "libpod/images/nonesuch/tag?repo=myrepo&tag=mytag" 404 # Tag the image -t POST "libpod/images/$IMAGE/tag?repo=localhost:5000/myrepo&tag=mytag" 201 +t POST "libpod/images/$IMAGE/tag?repo=localhost:$REGISTRY_PORT/myrepo&tag=mytag" 201 t GET libpod/images/$IMAGE/json 200 \ - .RepoTags[1]=localhost:5000/myrepo:mytag - -# Run registry container -# FIXME this fails if python tests have been run first... -podman run -d --name registry -p 5000:5000 quay.io/libpod/registry:2.7 /entrypoint.sh /etc/docker/registry/config.yml -wait_for_port localhost 5000 - -# Push to local registry and check output -while read -r LINE -do - if echo "${LINE}" | jq --exit-status 'select( .status != null) | select ( .status | contains("digest: sha256:"))' &>/dev/null; then - GOT_DIGEST="1" - fi -done < <(curl -sL "http://$HOST:$PORT/images/localhost:5000/myrepo/push?tlsVerify=false&tag=mytag" -XPOST) -if [ -z "${GOT_DIGEST}" ] ; then - echo -e "${red}not ok: did not found digest in output${nc}" 1>&2; -fi - -# Push to local registry -t POST "images/localhost:5000/myrepo/push?tlsVerify=false&tag=mytag" 200 + .RepoTags[1]=localhost:$REGISTRY_PORT/myrepo:mytag + +# Push to local registry... +t POST "images/localhost:$REGISTRY_PORT/myrepo/push?tlsVerify=false&tag=mytag" 200 + +# ...and check output. We can't use our built-in checks because this output +# is a sequence of JSON objects, i.e., individual ones, not in a JSON array. +# The lines themselves are valid JSON, but taken together they are not. +readarray lines <<<"$output" +s0=$(jq -r .status <<<"${lines[0]}") +is "$s0" "The push refers to repository [localhost:$REGISTRY_PORT/myrepo:mytag]" \ + "Push to local registry: first status line" + +# FIXME: is there a way to test the actual digest? +s1=$(jq -r .status <<<"${lines[1]}") +like "$s1" "mytag: digest: sha256:[0-9a-f]\{64\} size: [0-9]\+" \ + "Push to local registry: second status line" # Untag the image -t POST "libpod/images/$iid/untag?repo=localhost:5000/myrepo&tag=mytag" 201 +t POST "libpod/images/$iid/untag?repo=localhost:$REGISTRY_PORT/myrepo&tag=mytag" 201 # Try to push non-existing image -t POST "images/localhost:5000/idonotexist/push?tlsVerify=false" 404 +t POST "images/localhost:$REGISTRY_PORT/idonotexist/push?tlsVerify=false" 404 t GET libpod/images/$IMAGE/json 200 \ .RepoTags[-1]=$IMAGE -# Remove the registry container -t DELETE libpod/containers/registry?force=true 200 - -# Remove images +# Remove image t DELETE libpod/images/$IMAGE 200 \ .ExitCode=0 -t DELETE libpod/images/quay.io/libpod/registry:2.7 200 \ - .ExitCode=0 -if [ -z "${GOT_DIGEST}" ] ; then - exit 1; -fi +stop_registry diff --git a/test/apiv2/15-manifest.at b/test/apiv2/15-manifest.at index 0dd7026fa..970bed5a8 100644 --- a/test/apiv2/15-manifest.at +++ b/test/apiv2/15-manifest.at @@ -2,18 +2,40 @@ # # Tests for manifest list endpoints +start_registry + t POST /v3.4.0/libpod/manifests/create?name=abc 200 \ .Id~[0-9a-f]\\{64\\} id_abc=$(jq -r '.Id' <<<"$output") t POST /v4.0.0/libpod/manifests/xyz 201 \ .Id~[0-9a-f]\\{64\\} -echo xyz $output id_xyz=$(jq -r '.Id' <<<"$output") t GET /v3.4.0/libpod/manifests/$id_abc/exists 204 t GET /v4.0.0/libpod/manifests/$id_xyz/exists 204 +id_abc_image=$($PODMAN_BIN --root $WORKDIR/server_root image build -q --format=docker -<<EOF +FROM alpine +RUN >file1 +EOF +) + +id_xyz_image=$($PODMAN_BIN --root $WORKDIR/server_root image build -q --format=docker -<<EOF +FROM alpine +RUN >file2 +EOF +) + +t POST /v3.4.0/libpod/manifests/$id_abc/add images="[\"containers-storage:$id_abc_image\"]" 200 +t PUT /v4.0.0/libpod/manifests/$id_xyz operation='update' images="[\"containers-storage:$id_xyz_image\"]" 200 + +t POST "/v3.4.0/libpod/manifests/abc:latest/push?destination=localhost:$REGISTRY_PORT%2Fabc:latest&tlsVerify=false&all=true" 200 +t POST "/v4.0.0/libpod/manifests/xyz:latest/registry/localhost:$REGISTRY_PORT%2Fxyz:latest?tlsVerify=false&all=true" 200 + # /v3.x cannot delete a manifest list t DELETE /v4.0.0/libpod/manifests/$id_abc 200 t DELETE /v4.0.0/libpod/manifests/$id_xyz 200 + +podman rmi -a +stop_registry diff --git a/test/apiv2/20-containers.at b/test/apiv2/20-containers.at index 4d32a1031..383c527b4 100644 --- a/test/apiv2/20-containers.at +++ b/test/apiv2/20-containers.at @@ -45,16 +45,16 @@ t GET libpod/containers/json?all=true 200 \ .[0].IsInfra=false # Test compat API for Network Settings (.Network is N/A when rootless) -network_expect= +network_expect="Networks=null" if root; then - network_expect='.[0].NetworkSettings.Networks.podman.NetworkID=podman' + network_expect="Networks.podman.NetworkID=podman" fi t GET /containers/json?all=true 200 \ length=1 \ .[0].Id~[0-9a-f]\\{64\\} \ .[0].Image=$IMAGE \ .[0].Mounts~.*/tmp \ - $network_expect + .[0].NetworkSettings.$network_expect # compat API imageid with sha256: prefix t GET containers/json?limit=1 200 \ @@ -239,6 +239,7 @@ t GET containers/$cid/json 200 \ t POST containers/create Image=$IMAGE Entrypoint='["top"]' 201 \ .Id~[0-9a-f]\\{64\\} cid_top=$(jq -r '.Id' <<<"$output") + t GET containers/${cid_top}/json 200 \ .Config.Entrypoint[0]="top" \ .Config.Cmd='[]' \ @@ -477,7 +478,8 @@ for endpoint in containers/create libpod/containers/create; do t POST libpod/containers/$cid/init 204 - t GET libpod/containers/$cid/json 200 + t GET libpod/containers/$cid/json 200 \ + .HostsPath="" t DELETE containers/$cid 204 done diff --git a/test/apiv2/60-auth.at b/test/apiv2/60-auth.at index 1e087d12b..465b0a96d 100644 --- a/test/apiv2/60-auth.at +++ b/test/apiv2/60-auth.at @@ -3,7 +3,7 @@ # registry-related tests # -start_registry +start_registry htpasswd # Test unreachable t POST /v1.40/auth username=$REGISTRY_USERNAME password=WrOnGPassWord serveraddress=does.not.exist.io:1234/ \ @@ -26,3 +26,5 @@ t POST /v1.40/auth username=$REGISTRY_USERNAME password=$REGISTRY_PASSWORD serve 200 \ .Status="Login Succeeded" \ .IdentityToken="" + +stop_registry diff --git a/test/apiv2/70-short-names.at b/test/apiv2/70-short-names.at index dbf816f55..bd7f8e7bd 100644 --- a/test/apiv2/70-short-names.at +++ b/test/apiv2/70-short-names.at @@ -128,7 +128,7 @@ t DELETE "containers/$cid" # disable the docker.io enforcement. stop_service -CONTAINERS_CONF=$(pwd)/test/apiv2/containers.conf start_service +CONTAINERS_CONF=$TESTS_DIR/containers.conf start_service t POST "images/create?fromImage=quay.io/libpod/alpine:latest" 200 .error~null .status~".*Download complete.*" t POST "images/alpine/tag?repo=foo" 201 diff --git a/test/apiv2/test-apiv2 b/test/apiv2/test-apiv2 index c3545522e..6151b7672 100755 --- a/test/apiv2/test-apiv2 +++ b/test/apiv2/test-apiv2 @@ -62,7 +62,7 @@ clean_up_server() { podman rm -a podman rmi -af - stop_registry + stop_registry --cleanup stop_service fi } @@ -87,6 +87,7 @@ trap err_handler ERR ######### function die() { echo "$ME: $*" >&2 + clean_up_server exit 1 } @@ -219,19 +220,19 @@ function jsonify() { function t() { local method=$1; shift local path=$1; shift - local curl_args + local -a curl_args local content_type="application/json" local testname="$method $path" - # POST requests may be followed by one or more key=value pairs. + # POST and PUT requests may be followed by one or more key=value pairs. # Slurp the command line until we see a 3-digit status code. - if [[ $method = "POST" ]]; then + if [[ $method = "POST" || $method == "PUT" ]]; then local -a post_args for arg; do case "$arg" in *=*) post_args+=("$arg"); shift;; - *.tar) curl_args="--data-binary @$arg" ; + *.tar) curl_args+=(--data-binary @$arg); content_type="application/x-tar"; shift;; application/*) content_type="$arg"; @@ -241,8 +242,8 @@ function t() { esac done if [[ -z "$curl_args" ]]; then - curl_args="-d $(jsonify ${post_args[@]})" - testname="$testname [$curl_args]" + curl_args=(-d $(jsonify ${post_args[@]})) + testname="$testname [${curl_args[@]}]" fi fi @@ -269,7 +270,7 @@ function t() { # curl -X HEAD but without --head seems to wait for output anyway if [[ $method == "HEAD" ]]; then - curl_args="--head" + curl_args+=("--head") fi local expected_code=$1; shift @@ -281,7 +282,7 @@ function t() { # -s = silent, but --write-out 'format' gives us important response data # The hairy "{ ...;rc=$?; } || :" lets us capture curl's exit code and # give a helpful diagnostic if it fails. - { response=$(curl -s -X $method ${curl_args} \ + { response=$(curl -s -X $method "${curl_args[@]}" \ -H "Content-type: $content_type" \ --dump-header $WORKDIR/curl.headers.out \ --write-out '%{http_code}^%{content_type}^%{time_total}' \ @@ -289,8 +290,7 @@ function t() { # Any error from curl is instant bad news, from which we can't recover if [[ $rc -ne 0 ]]; then - echo "FATAL: curl failure ($rc) on $url - cannot continue" >&2 - exit 1 + die "curl failure ($rc) on $url - cannot continue" fi # Show returned headers (without trailing ^M or empty lines) in log file. @@ -380,11 +380,6 @@ function start_service() { die "Cannot start service on non-localhost ($HOST)" fi - echo "rootdir: "$WORKDIR - # Some tests use shortnames; force registry override to work around - # docker.io throttling. -# FIXME esm revisit pulling expected images re: shortnames caused tests to fail -# env CONTAINERS_REGISTRIES_CONF=$TESTS_DIR/../registries.conf $PODMAN_BIN \ --root $WORKDIR/server_root --syslog=true \ system service \ @@ -411,15 +406,17 @@ REGISTRY_PORT= REGISTRY_USERNAME= REGISTRY_PASSWORD= function start_registry() { - # We can be invoked multiple times, e.g. from different subtests, but - # let's assume that once started we only kill it at the end of tests. + # We can be called multiple times, but each time should start a new + # registry container with (possibly) different configuration. That + # means that all callers must be responsible for invoking stop_registry. if [[ -n "$REGISTRY_PORT" ]]; then - return + die "start_registry invoked twice in succession, without stop_registry" fi + # First arg is auth type (default: "none", but can also be "htpasswd") + local auth="${1:-none}" + REGISTRY_PORT=$(random_port) - REGISTRY_USERNAME=u$(random_string 7) - REGISTRY_PASSWORD=p$(random_string 7) local REGDIR=$WORKDIR/registry local AUTHDIR=$REGDIR/auth @@ -433,22 +430,33 @@ function start_registry() { podman ${PODMAN_REGISTRY_ARGS} pull $REGISTRY_IMAGE || podman ${PODMAN_REGISTRY_ARGS} pull $REGISTRY_IMAGE - # Create a local cert and credentials - # FIXME: is there a hidden "--quiet" flag? This is too noisy. - openssl req -newkey rsa:4096 -nodes -sha256 \ - -keyout $AUTHDIR/domain.key -x509 -days 2 \ - -out $AUTHDIR/domain.crt \ - -subj "/C=US/ST=Foo/L=Bar/O=Red Hat, Inc./CN=registry host certificate" \ - -addext subjectAltName=DNS:localhost - htpasswd -Bbn ${REGISTRY_USERNAME} ${REGISTRY_PASSWORD} \ - > $AUTHDIR/htpasswd + # Create a local cert (no need to do this more than once) + if [[ ! -e $AUTHDIR/domain.key ]]; then + # FIXME: is there a hidden "--quiet" flag? This is too noisy. + openssl req -newkey rsa:4096 -nodes -sha256 \ + -keyout $AUTHDIR/domain.key -x509 -days 2 \ + -out $AUTHDIR/domain.crt \ + -subj "/C=US/ST=Foo/L=Bar/O=Red Hat, Inc./CN=registry host certificate" \ + -addext subjectAltName=DNS:localhost + fi + + # If invoked with auth=htpasswd, create credentials + REGISTRY_USERNAME= + REGISTRY_PASSWORD= + if [[ "$auth" = "htpasswd" ]]; then + REGISTRY_USERNAME=u$(random_string 7) + REGISTRY_PASSWORD=p$(random_string 7) + + htpasswd -Bbn ${REGISTRY_USERNAME} ${REGISTRY_PASSWORD} \ + > $AUTHDIR/htpasswd + fi # Run the registry, and wait for it to come up podman ${PODMAN_REGISTRY_ARGS} run -d \ -p ${REGISTRY_PORT}:5000 \ --name registry \ -v $AUTHDIR:/auth:Z \ - -e "REGISTRY_AUTH=htpasswd" \ + -e "REGISTRY_AUTH=$auth" \ -e "REGISTRY_AUTH_HTPASSWD_REALM=Registry Realm" \ -e REGISTRY_AUTH_HTPASSWD_PATH=/auth/htpasswd \ -e REGISTRY_HTTP_TLS_CERTIFICATE=/auth/domain.crt \ @@ -462,13 +470,19 @@ function stop_registry() { local REGDIR=${WORKDIR}/registry if [[ -d $REGDIR ]]; then local OPTS="--root ${REGDIR}/root --runroot ${REGDIR}/runroot" - podman $OPTS stop -f -t 0 -a + podman $OPTS stop -i -t 0 registry # rm/rmi are important when running rootless: without them we # get EPERMS in tmpdir cleanup because files are owned by subuids. - podman $OPTS rm -f -a - podman $OPTS rmi -f -a + podman $OPTS rm -f -i registry + if [[ "$1" = "--cleanup" ]]; then + podman $OPTS rmi -f -a + fi fi + + REGISTRY_PORT= + REGISTRY_USERNAME= + REGISTRY_PASSWORD= } ################# diff --git a/test/e2e/pod_infra_container_test.go b/test/e2e/pod_infra_container_test.go index ad2db2411..20794a29c 100644 --- a/test/e2e/pod_infra_container_test.go +++ b/test/e2e/pod_infra_container_test.go @@ -130,14 +130,24 @@ var _ = Describe("Podman pod create", func() { session.WaitWithDefaultTimeout() Expect(session).Should(Exit(0)) - session = podmanTest.Podman([]string{"run", "-dt", "--pod", session.OutputToString(), ALPINE}) + session = podmanTest.Podman([]string{"run", "--name", "hostCtr", "--pod", session.OutputToString(), ALPINE, "readlink", "/proc/self/ns/net"}) session.WaitWithDefaultTimeout() Expect(session).Should(Exit(0)) - session = podmanTest.Podman([]string{"inspect", "--format", "'{{.NetworkSettings.SandboxKey}}'", session.OutputToString()}) + ns := SystemExec("readlink", []string{"/proc/self/ns/net"}) + ns.WaitWithDefaultTimeout() + Expect(ns).Should(Exit(0)) + netns := ns.OutputToString() + Expect(netns).ToNot(BeEmpty()) + + Expect(session.OutputToString()).To(Equal(netns)) + + // Sanity Check for podman inspect + session = podmanTest.Podman([]string{"inspect", "--format", "'{{.NetworkSettings.SandboxKey}}'", "hostCtr"}) session.WaitWithDefaultTimeout() Expect(session).Should(Exit(0)) - Expect(session.OutputToString()).Should(ContainSubstring("''")) // no network path... host + Expect(session.OutputToString()).Should(Equal("''")) // no network path... host + }) It("podman pod correctly sets up IPCNS", func() { diff --git a/test/system/170-run-userns.bats b/test/system/170-run-userns.bats index 46cb37b9d..84788a7f4 100644 --- a/test/system/170-run-userns.bats +++ b/test/system/170-run-userns.bats @@ -111,15 +111,30 @@ EOF } @test "podman userns=nomap" { - skip_if_not_rootless "--userns=nomap only works in rootless mode" - ns_user=$(id -un) - baseuid=$(egrep "${ns_user}:" /etc/subuid | cut -f2 -d:) - test ! -z ${baseuid} || skip "no IDs allocated for user ${ns_user}" + if is_rootless; then + ns_user=$(id -un) + baseuid=$(egrep "${ns_user}:" /etc/subuid | cut -f2 -d:) + test ! -z ${baseuid} || skip "no IDs allocated for user ${ns_user}" + + test_name="test_$(random_string 12)" + run_podman run -d --userns=nomap $IMAGE sleep 100 + cid=${output} + run_podman top ${cid} huser + is "${output}" "HUSER.*${baseuid}" "Container should start with baseuid from /etc/subuid not user UID" + run_podman rm -t 0 --force ${cid} + else + run_podman 125 run -d --userns=nomap $IMAGE sleep 100 + is "${output}" "Error: nomap is only supported in rootless mode" "Container should fail to start since nomap is not suppored in rootful mode" + fi +} - test_name="test_$(random_string 12)" - run_podman run -d --userns=nomap $IMAGE sleep 100 - cid=${output} - run_podman top ${cid} huser - is "${output}" "HUSER.*${baseuid}" "Container should start with baseuid from /etc/subuid not user UID" - run_podman rm -t 0 --force ${cid} +@test "podman userns=keep-id" { + if is_rootless; then + user=$(id -u) + run_podman run --rm --userns=keep-id $IMAGE id -u + is "${output}" "$user" "Container should run as the current user" + else + run_podman 125 run --rm --userns=keep-id $IMAGE id -u + is "${output}" "Error: keep-id is only supported in rootless mode" "Container should fail to start since keep-id is not suppored in rootful mode" + fi } diff --git a/test/system/520-checkpoint.bats b/test/system/520-checkpoint.bats index c16a8c35d..7f60f01b3 100644 --- a/test/system/520-checkpoint.bats +++ b/test/system/520-checkpoint.bats @@ -170,4 +170,34 @@ function teardown() { # FIXME: test --leave-running +@test "podman checkpoint --file-locks" { + action='flock test.lock sh -c "while [ -e /wait ];do sleep 0.5;done;for i in 1 2 3;do echo \$i;sleep 0.5;done"' + run_podman run -d $IMAGE sh -c "touch /wait; touch test.lock; echo READY; $action & $action & wait" + local cid="$output" + + # Wait for container to start emitting output + wait_for_ready $cid + + # Checkpoint, and confirm via inspect + run_podman container checkpoint --file-locks $cid + is "$output" "$cid" "podman container checkpoint" + + run_podman container inspect \ + --format '{{.State.Status}}:{{.State.Running}}:{{.State.Paused}}:{{.State.Checkpointed}}' $cid + is "$output" "exited:false:false:true" "State. Status:Running:Pause:Checkpointed" + + # Restart immediately and confirm state + run_podman container restore --file-locks $cid + is "$output" "$cid" "podman container restore" + + # Signal the container to continue; this is where the 1-2-3s will come from + run_podman exec $cid rm /wait + + # Wait for the container to stop + run_podman wait $cid + + run_podman logs $cid + trim=$(sed -z -e 's/[\r\n]\+//g' <<<"$output") + is "$trim" "READY123123" "File lock restored" +} # vim: filetype=sh diff --git a/test/system/600-completion.bats b/test/system/600-completion.bats index 2de9b1ae1..cb4a2c5f8 100644 --- a/test/system/600-completion.bats +++ b/test/system/600-completion.bats @@ -341,7 +341,9 @@ function _check_no_suggestions() { skip_if_remote "mounting via remote does not work" for cmd in create run; do run_completion $cmd $IMAGE "" - assert "$output" =~ ".*^/etc\$.*^/home\$.*^/root\$.*" "root directories suggested (cmd: podman $cmd)" + assert "$output" =~ ".*^/etc/\$.*" "etc directory suggested (cmd: podman $cmd)" + assert "$output" =~ ".*^/home/\$.*" "home directory suggested (cmd: podman $cmd)" + assert "$output" =~ ".*^/root/\$.*" "root directory suggested (cmd: podman $cmd)" # check completion for subdirectory run_completion $cmd $IMAGE "/etc" @@ -354,23 +356,31 @@ function _check_no_suggestions() { # check completion with relative path components # It is important the we will still use the image root and not escape to the host run_completion $cmd $IMAGE "../../" - assert "$output" =~ ".*^../../etc\$.*^../../home\$.*" "relative root directories suggested (cmd: podman $cmd ../../)" + assert "$output" =~ ".*^../../etc/\$.*" "relative etc directory suggested (cmd: podman $cmd ../../)" + assert "$output" =~ ".*^../../home/\$.*" "relative home directory suggested (cmd: podman $cmd ../../)" done random_name=$(random_string 30) random_file=$(random_string 30) - run_podman run --name $random_name $IMAGE touch /tmp/$random_file + run_podman run --name $random_name $IMAGE sh -c "touch /tmp/$random_file && touch /tmp/${random_file}2 && mkdir /emptydir" # check completion for podman cp run_completion cp "" assert "$output" =~ ".*^$random_name\:\$.*" "podman cp suggest container names" run_completion cp "$random_name:" - assert "$output" =~ ".*^$random_name\:/etc\$.*" "podman cp suggest paths in container" + assert "$output" =~ ".*^$random_name\:/etc/\$.*" "podman cp suggest paths in container" run_completion cp "$random_name:/tmp" assert "$output" =~ ".*^$random_name\:/tmp/$random_file\$.*" "podman cp suggest custom file in container" + run_completion cp "$random_name:/tmp/$random_file" + assert "$output" =~ ".*^$random_name\:/tmp/$random_file\$.*" "podman cp suggest /tmp/$random_file file in container" + assert "$output" =~ ".*^$random_name\:/tmp/${random_file}2\$.*" "podman cp suggest /tmp/${random_file}2 file in container" + + run_completion cp "$random_name:/emptydir" + assert "$output" =~ ".*^$random_name\:/emptydir/\$.*ShellCompDirectiveNoSpace" "podman cp suggest empty dir with no space directive (:2)" + # cleanup container run_podman rm $random_name } diff --git a/vendor/github.com/seccomp/libseccomp-golang/.golangci.yml b/vendor/github.com/seccomp/libseccomp-golang/.golangci.yml new file mode 100644 index 000000000..7df8aa198 --- /dev/null +++ b/vendor/github.com/seccomp/libseccomp-golang/.golangci.yml @@ -0,0 +1,4 @@ +# For documentation, see https://golangci-lint.run/usage/configuration/ +linters: + enable: + - gofumpt diff --git a/vendor/github.com/seccomp/libseccomp-golang/.travis.yml b/vendor/github.com/seccomp/libseccomp-golang/.travis.yml deleted file mode 100644 index 5240d4622..000000000 --- a/vendor/github.com/seccomp/libseccomp-golang/.travis.yml +++ /dev/null @@ -1,57 +0,0 @@ -# Travis CI configuration for libseccomp-golang - -# https://docs.travis-ci.com/user/reference/bionic -# https://wiki.ubuntu.com/Releases - -dist: bionic -sudo: false - -notifications: - email: - on_success: always - on_failure: always - -arch: - - amd64 - -os: - - linux - -language: go - -jobs: - include: - - name: "last libseccomp 2.5.0" - env: - - SECCOMP_VER=2.5.0 - - SECCOMP_SHA256SUM=1ffa7038d2720ad191919816db3479295a4bcca1ec14e02f672539f4983014f3 - - name: "compat libseccomp 2.4.4" - env: - - SECCOMP_VER=2.4.4 - - SECCOMP_SHA256SUM=4e79738d1ef3c9b7ca9769f1f8b8d84fc17143c2c1c432e53b9c64787e0ff3eb - - name: "compat libseccomp 2.2.1" - env: - - SECCOMP_VER=2.2.1 - - SECCOMP_SHA256SUM=0ba1789f54786c644af54cdffc9fd0dd0a8bb2b2ee153933f658855d2851a740 - -addons: - apt: - packages: - - build-essential - - astyle - - golint - - gperf - -install: - - go get -u golang.org/x/lint/golint - -# run all of the tests independently, fail if any of the tests error -script: - - wget https://github.com/seccomp/libseccomp/releases/download/v$SECCOMP_VER/libseccomp-$SECCOMP_VER.tar.gz - - echo $SECCOMP_SHA256SUM libseccomp-$SECCOMP_VER.tar.gz | sha256sum -c - - tar xf libseccomp-$SECCOMP_VER.tar.gz - - pushd libseccomp-$SECCOMP_VER && ./configure --prefix=/opt/libseccomp-$SECCOMP_VER && make && sudo make install && popd - - make check-syntax - - make lint - - PKG_CONFIG_PATH=/opt/libseccomp-$SECCOMP_VER/lib/pkgconfig LD_LIBRARY_PATH=/opt/libseccomp-$SECCOMP_VER/lib make vet - - PKG_CONFIG_PATH=/opt/libseccomp-$SECCOMP_VER/lib/pkgconfig LD_LIBRARY_PATH=/opt/libseccomp-$SECCOMP_VER/lib make test diff --git a/vendor/github.com/seccomp/libseccomp-golang/CONTRIBUTING.md b/vendor/github.com/seccomp/libseccomp-golang/CONTRIBUTING.md index d6862cbd5..c2fc80d5a 100644 --- a/vendor/github.com/seccomp/libseccomp-golang/CONTRIBUTING.md +++ b/vendor/github.com/seccomp/libseccomp-golang/CONTRIBUTING.md @@ -1,31 +1,23 @@ -How to Submit Patches to the libseccomp Project +How to Submit Patches to the libseccomp-golang Project =============================================================================== https://github.com/seccomp/libseccomp-golang This document is intended to act as a guide to help you contribute to the -libseccomp project. It is not perfect, and there will always be exceptions -to the rules described here, but by following the instructions below you -should have a much easier time getting your work merged with the upstream +libseccomp-golang project. It is not perfect, and there will always be +exceptions to the rules described here, but by following the instructions below +you should have a much easier time getting your work merged with the upstream project. ## Test Your Code Using Existing Tests -There are two possible tests you can run to verify your code. The first -test is used to check the formatting and coding style of your changes, you -can run the test with the following command: - - # make check-syntax - -... if there are any problems with your changes a diff/patch will be shown -which indicates the problems and how to fix them. - -The second possible test is used to ensure the sanity of your code changes -and to test these changes against the included tests. You can run the test -with the following command: +A number of tests and lint related recipes are provided in the Makefile, if +you want to run the standard regression tests, you can execute the following: # make check -... if there are any faults or errors they will be displayed. +In order to use it, the 'golangci-lint' tool is needed, which can be found at: + +* https://github.com/golangci/golangci-lint ## Add New Tests for New Functionality diff --git a/vendor/github.com/seccomp/libseccomp-golang/Makefile b/vendor/github.com/seccomp/libseccomp-golang/Makefile index 38cfa852c..530f5b4ad 100644 --- a/vendor/github.com/seccomp/libseccomp-golang/Makefile +++ b/vendor/github.com/seccomp/libseccomp-golang/Makefile @@ -4,7 +4,7 @@ all: check-build -check: vet test +check: lint test check-build: go build @@ -16,7 +16,7 @@ fix-syntax: gofmt -w . vet: - go vet -v + go vet -v ./... # Previous bugs have made the tests freeze until the timeout. Golang default # timeout for tests is 10 minutes, which is too long, considering current tests @@ -28,5 +28,4 @@ test: go test -v -timeout $(TEST_TIMEOUT) lint: - @$(if $(shell which golint),true,$(error "install golint and include it in your PATH")) - golint -set_exit_status + golangci-lint run . diff --git a/vendor/github.com/seccomp/libseccomp-golang/README.md b/vendor/github.com/seccomp/libseccomp-golang/README.md index 806a5ddf2..6430f1c9e 100644 --- a/vendor/github.com/seccomp/libseccomp-golang/README.md +++ b/vendor/github.com/seccomp/libseccomp-golang/README.md @@ -2,7 +2,9 @@ =============================================================================== https://github.com/seccomp/libseccomp-golang -[](https://travis-ci.org/seccomp/libseccomp-golang) +[](https://pkg.go.dev/github.com/seccomp/libseccomp-golang) +[](https://github.com/seccomp/libseccomp-golang/actions/workflows/validate.yml) +[](https://github.com/seccomp/libseccomp-golang/actions/workflows/test.yml) The libseccomp library provides an easy to use, platform independent, interface to the Linux Kernel's syscall filtering mechanism. The libseccomp API is @@ -26,26 +28,14 @@ list. * https://groups.google.com/d/forum/libseccomp -Documentation is also available at: +Documentation for this package is also available at: -* https://godoc.org/github.com/seccomp/libseccomp-golang +* https://pkg.go.dev/github.com/seccomp/libseccomp-golang ## Installing the package -The libseccomp-golang bindings require at least Go v1.2.1 and GCC v4.8.4; -earlier versions may yield unpredictable results. If you meet these -requirements you can install this package using the command below: - # go get github.com/seccomp/libseccomp-golang -## Testing the Library - -A number of tests and lint related recipes are provided in the Makefile, if -you want to run the standard regression tests, you can excute the following: - - # make check - -In order to execute the 'make lint' recipe the 'golint' tool is needed, it -can be found at: +## Contributing -* https://github.com/golang/lint +See [CONTRIBUTING.md](CONTRIBUTING.md). diff --git a/vendor/github.com/seccomp/libseccomp-golang/SECURITY.md b/vendor/github.com/seccomp/libseccomp-golang/SECURITY.md new file mode 100644 index 000000000..c448faa8e --- /dev/null +++ b/vendor/github.com/seccomp/libseccomp-golang/SECURITY.md @@ -0,0 +1,47 @@ +The libseccomp-golang Security Vulnerability Handling Process +=============================================================================== +https://github.com/seccomp/libseccomp-golang + +This document document attempts to describe the processes through which +sensitive security relevant bugs can be responsibly disclosed to the +libseccomp-golang project and how the project maintainers should handle these +reports. Just like the other libseccomp-golang process documents, this +document should be treated as a guiding document and not a hard, unyielding set +of regulations; the bug reporters and project maintainers are encouraged to +work together to address the issues as best they can, in a manner which works +best for all parties involved. + +### Reporting Problems + +Problems with the libseccomp-golang library that are not suitable for immediate +public disclosure should be emailed to the current libseccomp-golang +maintainers, the list is below. We typically request at most a 90 day time +period to address the issue before it is made public, but we will make every +effort to address the issue as quickly as possible and shorten the disclosure +window. + +* Paul Moore, paul@paul-moore.com +* Tom Hromatka, tom.hromatka@oracle.com + +### Resolving Sensitive Security Issues + +Upon disclosure of a bug, the maintainers should work together to investigate +the problem and decide on a solution. In order to prevent an early disclosure +of the problem, those working on the solution should do so privately and +outside of the traditional libseccomp-golang development practices. One +possible solution to this is to leverage the GitHub "Security" functionality to +create a private development fork that can be shared among the maintainers, and +optionally the reporter. A placeholder GitHub issue may be created, but +details should remain extremely limited until such time as the problem has been +fixed and responsibly disclosed. If a CVE, or other tag, has been assigned to +the problem, the GitHub issue title should include the vulnerability tag once +the problem has been disclosed. + +### Public Disclosure + +Whenever possible, responsible reporting and patching practices should be +followed, including notification to the linux-distros and oss-security mailing +lists. + +* https://oss-security.openwall.org/wiki/mailing-lists/distros +* https://oss-security.openwall.org/wiki/mailing-lists/oss-security diff --git a/vendor/github.com/seccomp/libseccomp-golang/go.sum b/vendor/github.com/seccomp/libseccomp-golang/go.sum index 72ae16111..e69de29bb 100644 --- a/vendor/github.com/seccomp/libseccomp-golang/go.sum +++ b/vendor/github.com/seccomp/libseccomp-golang/go.sum @@ -1,23 +0,0 @@ -github.com/yuin/goldmark v1.1.25/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74= -golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w= -golang.org/x/crypto v0.0.0-20191011191535-87dc89f01550/go.mod h1:yigFU9vqHzYiE8UmvKecakEJjdnWj3jj499lnFckfCI= -golang.org/x/lint v0.0.0-20200302205851-738671d3881b h1:Wh+f8QHJXR411sJR8/vRBTZ7YapZaRvUcLFFJhusH0k= -golang.org/x/lint v0.0.0-20200302205851-738671d3881b/go.mod h1:3xt1FjdF8hUf6vQPIChWIBhFzV8gjjsPE/fR3IyQdNY= -golang.org/x/mod v0.1.1-0.20191105210325-c90efee705ee/go.mod h1:QqPTAvyqsEbceGzBzNggFXnrqF1CaUcvgkdR5Ot7KZg= -golang.org/x/mod v0.2.0/go.mod h1:s0Qsj1ACt9ePp/hMypM3fl4fZqREWJwdYDEqhRiZZUA= -golang.org/x/net v0.0.0-20190404232315-eb5bcb51f2a3/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg= -golang.org/x/net v0.0.0-20190620200207-3b0461eec859/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s= -golang.org/x/net v0.0.0-20200226121028-0de0cce0169b/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s= -golang.org/x/sync v0.0.0-20190423024810-112230192c58/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM= -golang.org/x/sync v0.0.0-20190911185100-cd5d95a43a6e/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM= -golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY= -golang.org/x/sys v0.0.0-20190412213103-97732733099d/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs= -golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ= -golang.org/x/tools v0.0.0-20191119224855-298f0cb1881e/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo= -golang.org/x/tools v0.0.0-20200130002326-2f3ba24bd6e7 h1:EBZoQjiKKPaLbPrbpssUfuHtwM6KV/vb4U85g/cigFY= -golang.org/x/tools v0.0.0-20200130002326-2f3ba24bd6e7/go.mod h1:TB2adYChydJhpapKDTa4BR/hXlZSLoq2Wpct/0txZ28= -golang.org/x/tools v0.0.0-20200313205530-4303120df7d8 h1:gkI/wGGwpcG5W4hLCzZNGxA4wzWBGGDStRI1MrjDl2Q= -golang.org/x/tools v0.0.0-20200313205530-4303120df7d8/go.mod h1:Sl4aGygMT6LrqrWclx+PTx3U+LnKx/seiNR+3G19Ar8= -golang.org/x/xerrors v0.0.0-20190717185122-a985d3407aa7/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0= -golang.org/x/xerrors v0.0.0-20191011141410-1b5146add898/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0= -golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0= diff --git a/vendor/github.com/seccomp/libseccomp-golang/seccomp.go b/vendor/github.com/seccomp/libseccomp-golang/seccomp.go index e9b92e221..8dad12fdb 100644 --- a/vendor/github.com/seccomp/libseccomp-golang/seccomp.go +++ b/vendor/github.com/seccomp/libseccomp-golang/seccomp.go @@ -1,5 +1,3 @@ -// +build linux - // Public API specification for libseccomp Go bindings // Contains public API for the bindings @@ -18,48 +16,36 @@ import ( "unsafe" ) -// C wrapping code - -// To compile libseccomp-golang against a specific version of libseccomp: -// cd ../libseccomp && mkdir -p prefix -// ./configure --prefix=$PWD/prefix && make && make install -// cd ../libseccomp-golang -// PKG_CONFIG_PATH=$PWD/../libseccomp/prefix/lib/pkgconfig/ make -// LD_PRELOAD=$PWD/../libseccomp/prefix/lib/libseccomp.so.2.5.0 PKG_CONFIG_PATH=$PWD/../libseccomp/prefix/lib/pkgconfig/ make test - -// #cgo pkg-config: libseccomp // #include <stdlib.h> // #include <seccomp.h> import "C" // Exported types -// VersionError denotes that the system libseccomp version is incompatible -// with this package. +// VersionError represents an error when either the system libseccomp version +// or the kernel version is too old to perform the operation requested. type VersionError struct { - message string - minimum string + op string // operation that failed or would fail + major, minor, micro uint // minimally required libseccomp version + curAPI, minAPI uint // current and minimally required API versions } func init() { // This forces the cgo libseccomp to initialize its internal API support state, // which is necessary on older versions of libseccomp in order to work // correctly. - GetAPI() + _, _ = getAPI() } func (e VersionError) Error() string { - messageStr := "" - if e.message != "" { - messageStr = e.message + ": " + if e.minAPI != 0 { + return fmt.Sprintf("%s requires libseccomp >= %d.%d.%d and API level >= %d "+ + "(current version: %d.%d.%d, API level: %d)", + e.op, e.major, e.minor, e.micro, e.minAPI, + verMajor, verMinor, verMicro, e.curAPI) } - minimumStr := "" - if e.minimum != "" { - minimumStr = e.minimum - } else { - minimumStr = "2.2.0" - } - return fmt.Sprintf("Libseccomp version too low: %sminimum supported is %s: detected %d.%d.%d", messageStr, minimumStr, verMajor, verMinor, verMicro) + return fmt.Sprintf("%s requires libseccomp >= %d.%d.%d (current version: %d.%d.%d)", + e.op, e.major, e.minor, e.micro, verMajor, verMinor, verMicro) } // ScmpArch represents a CPU architecture. Seccomp can restrict syscalls on a @@ -148,44 +134,46 @@ const ( // variables are invalid ArchInvalid ScmpArch = iota // ArchNative is the native architecture of the kernel - ArchNative ScmpArch = iota + ArchNative // ArchX86 represents 32-bit x86 syscalls - ArchX86 ScmpArch = iota + ArchX86 // ArchAMD64 represents 64-bit x86-64 syscalls - ArchAMD64 ScmpArch = iota + ArchAMD64 // ArchX32 represents 64-bit x86-64 syscalls (32-bit pointers) - ArchX32 ScmpArch = iota + ArchX32 // ArchARM represents 32-bit ARM syscalls - ArchARM ScmpArch = iota + ArchARM // ArchARM64 represents 64-bit ARM syscalls - ArchARM64 ScmpArch = iota + ArchARM64 // ArchMIPS represents 32-bit MIPS syscalls - ArchMIPS ScmpArch = iota + ArchMIPS // ArchMIPS64 represents 64-bit MIPS syscalls - ArchMIPS64 ScmpArch = iota + ArchMIPS64 // ArchMIPS64N32 represents 64-bit MIPS syscalls (32-bit pointers) - ArchMIPS64N32 ScmpArch = iota + ArchMIPS64N32 // ArchMIPSEL represents 32-bit MIPS syscalls (little endian) - ArchMIPSEL ScmpArch = iota + ArchMIPSEL // ArchMIPSEL64 represents 64-bit MIPS syscalls (little endian) - ArchMIPSEL64 ScmpArch = iota + ArchMIPSEL64 // ArchMIPSEL64N32 represents 64-bit MIPS syscalls (little endian, // 32-bit pointers) - ArchMIPSEL64N32 ScmpArch = iota + ArchMIPSEL64N32 // ArchPPC represents 32-bit POWERPC syscalls - ArchPPC ScmpArch = iota + ArchPPC // ArchPPC64 represents 64-bit POWER syscalls (big endian) - ArchPPC64 ScmpArch = iota + ArchPPC64 // ArchPPC64LE represents 64-bit POWER syscalls (little endian) - ArchPPC64LE ScmpArch = iota + ArchPPC64LE // ArchS390 represents 31-bit System z/390 syscalls - ArchS390 ScmpArch = iota + ArchS390 // ArchS390X represents 64-bit System z/390 syscalls - ArchS390X ScmpArch = iota + ArchS390X // ArchPARISC represents 32-bit PA-RISC - ArchPARISC ScmpArch = iota + ArchPARISC // ArchPARISC64 represents 64-bit PA-RISC - ArchPARISC64 ScmpArch = iota + ArchPARISC64 + // ArchRISCV64 represents RISCV64 + ArchRISCV64 ) const ( @@ -194,34 +182,36 @@ const ( // ActInvalid is a placeholder to ensure uninitialized ScmpAction // variables are invalid ActInvalid ScmpAction = iota - // ActKill kills the thread that violated the rule. It is the same as ActKillThread. + // ActKillThread kills the thread that violated the rule. // All other threads from the same thread group will continue to execute. - ActKill ScmpAction = iota + ActKillThread // ActTrap throws SIGSYS - ActTrap ScmpAction = iota + ActTrap // ActNotify triggers a userspace notification. This action is only usable when // libseccomp API level 6 or higher is supported. - ActNotify ScmpAction = iota + ActNotify // ActErrno causes the syscall to return a negative error code. This // code can be set with the SetReturnCode method - ActErrno ScmpAction = iota + ActErrno // ActTrace causes the syscall to notify tracing processes with the // given error code. This code can be set with the SetReturnCode method - ActTrace ScmpAction = iota + ActTrace // ActAllow permits the syscall to continue execution - ActAllow ScmpAction = iota + ActAllow // ActLog permits the syscall to continue execution after logging it. // This action is only usable when libseccomp API level 3 or higher is // supported. - ActLog ScmpAction = iota - // ActKillThread kills the thread that violated the rule. It is the same as ActKill. - // All other threads from the same thread group will continue to execute. - ActKillThread ScmpAction = iota + ActLog // ActKillProcess kills the process that violated the rule. // All threads in the thread group are also terminated. // This action is only usable when libseccomp API level 3 or higher is // supported. - ActKillProcess ScmpAction = iota + ActKillProcess + // ActKill kills the thread that violated the rule. + // All other threads from the same thread group will continue to execute. + // + // Deprecated: use ActKillThread + ActKill = ActKillThread ) const ( @@ -234,36 +224,35 @@ const ( CompareInvalid ScmpCompareOp = iota // CompareNotEqual returns true if the argument is not equal to the // given value - CompareNotEqual ScmpCompareOp = iota + CompareNotEqual // CompareLess returns true if the argument is less than the given value - CompareLess ScmpCompareOp = iota + CompareLess // CompareLessOrEqual returns true if the argument is less than or equal // to the given value - CompareLessOrEqual ScmpCompareOp = iota + CompareLessOrEqual // CompareEqual returns true if the argument is equal to the given value - CompareEqual ScmpCompareOp = iota + CompareEqual // CompareGreaterEqual returns true if the argument is greater than or // equal to the given value - CompareGreaterEqual ScmpCompareOp = iota + CompareGreaterEqual // CompareGreater returns true if the argument is greater than the given // value - CompareGreater ScmpCompareOp = iota - // CompareMaskedEqual returns true if the argument is equal to the given - // value, when masked (bitwise &) against the second given value - CompareMaskedEqual ScmpCompareOp = iota + CompareGreater + // CompareMaskedEqual returns true if the masked argument value is + // equal to the masked datum value. Mask is the first argument, and + // datum is the second one. + CompareMaskedEqual ) -var ( - // ErrSyscallDoesNotExist represents an error condition where - // libseccomp is unable to resolve the syscall - ErrSyscallDoesNotExist = fmt.Errorf("could not resolve syscall name") -) +// ErrSyscallDoesNotExist represents an error condition where +// libseccomp is unable to resolve the syscall +var ErrSyscallDoesNotExist = fmt.Errorf("could not resolve syscall name") const ( // Userspace notification response flags // NotifRespFlagContinue tells the kernel to continue executing the system - // call that triggered the notification. Must only be used when the notication + // call that triggered the notification. Must only be used when the notification // response's error is 0. NotifRespFlagContinue uint32 = 1 ) @@ -314,6 +303,8 @@ func GetArchFromString(arch string) (ScmpArch, error) { return ArchPARISC, nil case "parisc64": return ArchPARISC64, nil + case "riscv64": + return ArchRISCV64, nil default: return ArchInvalid, fmt.Errorf("cannot convert unrecognized string %q", arch) } @@ -358,6 +349,8 @@ func (a ScmpArch) String() string { return "parisc" case ArchPARISC64: return "parisc64" + case ArchRISCV64: + return "riscv64" case ArchNative: return "native" case ArchInvalid: @@ -394,7 +387,7 @@ func (a ScmpCompareOp) String() string { // String returns a string representation of a seccomp match action func (a ScmpAction) String() string { switch a & 0xFFFF { - case ActKill, ActKillThread: + case ActKillThread: return "Action: Kill thread" case ActKillProcess: return "Action: Kill process" @@ -556,8 +549,8 @@ func MakeCondition(arg uint, comparison ScmpCompareOp, values ...uint64) (ScmpCo return condStruct, err } - if comparison == CompareInvalid { - return condStruct, fmt.Errorf("invalid comparison operator") + if err := sanitizeCompareOp(comparison); err != nil { + return condStruct, err } else if arg > 5 { return condStruct, fmt.Errorf("syscalls only have up to 6 arguments (%d given)", arg) } else if len(values) > 2 { @@ -874,10 +867,8 @@ func (f *ScmpFilter) GetNoNewPrivsBit() (bool, error) { func (f *ScmpFilter) GetLogBit() (bool, error) { log, err := f.getFilterAttr(filterAttrLog) if err != nil { - // Ignore error, if not supported returns apiLevel == 0 - apiLevel, _ := GetAPI() - if apiLevel < 3 { - return false, fmt.Errorf("getting the log bit is only supported in libseccomp 2.4.0 and newer with API level 3 or higher") + if e := checkAPI("GetLogBit", 3, 2, 4, 0); e != nil { + err = e } return false, err @@ -899,9 +890,8 @@ func (f *ScmpFilter) GetLogBit() (bool, error) { func (f *ScmpFilter) GetSSB() (bool, error) { ssb, err := f.getFilterAttr(filterAttrSSB) if err != nil { - api, apiErr := getAPI() - if (apiErr != nil && api == 0) || (apiErr == nil && api < 4) { - return false, fmt.Errorf("getting the SSB flag is only supported in libseccomp 2.5.0 and newer with API level 4 or higher") + if e := checkAPI("GetSSB", 4, 2, 5, 0); e != nil { + err = e } return false, err @@ -914,6 +904,42 @@ func (f *ScmpFilter) GetSSB() (bool, error) { return true, nil } +// GetOptimize returns the current optimization level of the filter, +// or an error if an issue was encountered retrieving the value. +// See SetOptimize for more details. +func (f *ScmpFilter) GetOptimize() (int, error) { + level, err := f.getFilterAttr(filterAttrOptimize) + if err != nil { + if e := checkAPI("GetOptimize", 4, 2, 5, 0); e != nil { + err = e + } + + return 0, err + } + + return int(level), nil +} + +// GetRawRC returns the current state of RawRC flag, or an error +// if an issue was encountered retrieving the value. +// See SetRawRC for more details. +func (f *ScmpFilter) GetRawRC() (bool, error) { + rawrc, err := f.getFilterAttr(filterAttrRawRC) + if err != nil { + if e := checkAPI("GetRawRC", 4, 2, 5, 0); e != nil { + err = e + } + + return false, err + } + + if rawrc == 0 { + return false, nil + } + + return true, nil +} + // SetBadArchAction sets the default action taken on a syscall for an // architecture not in the filter, or an error if an issue was encountered // setting the value. @@ -953,10 +979,8 @@ func (f *ScmpFilter) SetLogBit(state bool) error { err := f.setFilterAttr(filterAttrLog, toSet) if err != nil { - // Ignore error, if not supported returns apiLevel == 0 - apiLevel, _ := GetAPI() - if apiLevel < 3 { - return fmt.Errorf("setting the log bit is only supported in libseccomp 2.4.0 and newer with API level 3 or higher") + if e := checkAPI("SetLogBit", 3, 2, 4, 0); e != nil { + err = e } } @@ -976,9 +1000,52 @@ func (f *ScmpFilter) SetSSB(state bool) error { err := f.setFilterAttr(filterAttrSSB, toSet) if err != nil { - api, apiErr := getAPI() - if (apiErr != nil && api == 0) || (apiErr == nil && api < 4) { - return fmt.Errorf("setting the SSB flag is only supported in libseccomp 2.5.0 and newer with API level 4 or higher") + if e := checkAPI("SetSSB", 4, 2, 5, 0); e != nil { + err = e + } + } + + return err +} + +// SetOptimize sets optimization level of the seccomp filter. By default +// libseccomp generates a set of sequential "if" statements for each rule in +// the filter. SetSyscallPriority can be used to prioritize the order for the +// default cause. The binary tree optimization sorts by syscall numbers and +// generates consistent O(log n) filter traversal for every rule in the filter. +// The binary tree may be advantageous for large filters. Note that +// SetSyscallPriority is ignored when level == 2. +// +// The different optimization levels are: +// 0: Reserved value, not currently used. +// 1: Rules sorted by priority and complexity (DEFAULT). +// 2: Binary tree sorted by syscall number. +func (f *ScmpFilter) SetOptimize(level int) error { + cLevel := C.uint32_t(level) + + err := f.setFilterAttr(filterAttrOptimize, cLevel) + if err != nil { + if e := checkAPI("SetOptimize", 4, 2, 5, 0); e != nil { + err = e + } + } + + return err +} + +// SetRawRC sets whether libseccomp should pass system error codes back to the +// caller, instead of the default ECANCELED. Defaults to false. +func (f *ScmpFilter) SetRawRC(state bool) error { + var toSet C.uint32_t = 0x0 + + if state { + toSet = 0x1 + } + + err := f.setFilterAttr(filterAttrRawRC, toSet) + if err != nil { + if e := checkAPI("SetRawRC", 4, 2, 5, 0); e != nil { + err = e } } @@ -1029,9 +1096,6 @@ func (f *ScmpFilter) AddRuleExact(call ScmpSyscall, action ScmpAction) error { // AddRuleConditional adds a single rule for a conditional action on a syscall. // Returns an error if an issue was encountered adding the rule. // All conditions must match for the rule to match. -// There is a bug in library versions below v2.2.1 which can, in some cases, -// cause conditions to be lost when more than one are used. Consequently, -// AddRuleConditional is disabled on library versions lower than v2.2.1 func (f *ScmpFilter) AddRuleConditional(call ScmpSyscall, action ScmpAction, conds []ScmpCondition) error { return f.addRuleGeneric(call, action, false, conds) } @@ -1043,9 +1107,6 @@ func (f *ScmpFilter) AddRuleConditional(call ScmpSyscall, action ScmpAction, con // The rule will function exactly as described, but it may not function identically // (or be able to be applied to) all architectures. // Returns an error if an issue was encountered adding the rule. -// There is a bug in library versions below v2.2.1 which can, in some cases, -// cause conditions to be lost when more than one are used. Consequently, -// AddRuleConditionalExact is disabled on library versions lower than v2.2.1 func (f *ScmpFilter) AddRuleConditionalExact(call ScmpSyscall, action ScmpAction, conds []ScmpCondition) error { return f.addRuleGeneric(call, action, true, conds) } diff --git a/vendor/github.com/seccomp/libseccomp-golang/seccomp_internal.go b/vendor/github.com/seccomp/libseccomp-golang/seccomp_internal.go index 8dc7b296f..df4dfb7eb 100644 --- a/vendor/github.com/seccomp/libseccomp-golang/seccomp_internal.go +++ b/vendor/github.com/seccomp/libseccomp-golang/seccomp_internal.go @@ -1,11 +1,10 @@ -// +build linux - // Internal functions for libseccomp Go bindings // No exported functions package seccomp import ( + "errors" "fmt" "syscall" ) @@ -27,10 +26,10 @@ import ( #include <stdlib.h> #include <seccomp.h> -#if SCMP_VER_MAJOR < 2 -#error Minimum supported version of Libseccomp is v2.2.0 -#elif SCMP_VER_MAJOR == 2 && SCMP_VER_MINOR < 2 -#error Minimum supported version of Libseccomp is v2.2.0 +#if (SCMP_VER_MAJOR < 2) || \ + (SCMP_VER_MAJOR == 2 && SCMP_VER_MINOR < 3) || \ + (SCMP_VER_MAJOR == 2 && SCMP_VER_MINOR == 3 && SCMP_VER_MICRO < 1) +#error This package requires libseccomp >= v2.3.1 #endif #define ARCH_BAD ~0 @@ -65,6 +64,10 @@ const uint32_t C_ARCH_BAD = ARCH_BAD; #define SCMP_ARCH_PARISC64 ARCH_BAD #endif +#ifndef SCMP_ARCH_RISCV64 +#define SCMP_ARCH_RISCV64 ARCH_BAD +#endif + const uint32_t C_ARCH_NATIVE = SCMP_ARCH_NATIVE; const uint32_t C_ARCH_X86 = SCMP_ARCH_X86; const uint32_t C_ARCH_X86_64 = SCMP_ARCH_X86_64; @@ -84,6 +87,7 @@ const uint32_t C_ARCH_S390 = SCMP_ARCH_S390; const uint32_t C_ARCH_S390X = SCMP_ARCH_S390X; const uint32_t C_ARCH_PARISC = SCMP_ARCH_PARISC; const uint32_t C_ARCH_PARISC64 = SCMP_ARCH_PARISC64; +const uint32_t C_ARCH_RISCV64 = SCMP_ARCH_RISCV64; #ifndef SCMP_ACT_LOG #define SCMP_ACT_LOG 0x7ffc0000U @@ -113,20 +117,25 @@ const uint32_t C_ACT_NOTIFY = SCMP_ACT_NOTIFY; // The libseccomp SCMP_FLTATR_CTL_LOG member of the scmp_filter_attr enum was // added in v2.4.0 -#if (SCMP_VER_MAJOR < 2) || \ - (SCMP_VER_MAJOR == 2 && SCMP_VER_MINOR < 4) +#if SCMP_VER_MAJOR == 2 && SCMP_VER_MINOR < 4 #define SCMP_FLTATR_CTL_LOG _SCMP_FLTATR_MIN #endif + +// The following SCMP_FLTATR_* were added in libseccomp v2.5.0. #if SCMP_VER_MAJOR == 2 && SCMP_VER_MINOR < 5 -#define SCMP_FLTATR_CTL_SSB _SCMP_FLTATR_MIN +#define SCMP_FLTATR_CTL_SSB _SCMP_FLTATR_MIN +#define SCMP_FLTATR_CTL_OPTIMIZE _SCMP_FLTATR_MIN +#define SCMP_FLTATR_API_SYSRAWRC _SCMP_FLTATR_MIN #endif -const uint32_t C_ATTRIBUTE_DEFAULT = (uint32_t)SCMP_FLTATR_ACT_DEFAULT; -const uint32_t C_ATTRIBUTE_BADARCH = (uint32_t)SCMP_FLTATR_ACT_BADARCH; -const uint32_t C_ATTRIBUTE_NNP = (uint32_t)SCMP_FLTATR_CTL_NNP; -const uint32_t C_ATTRIBUTE_TSYNC = (uint32_t)SCMP_FLTATR_CTL_TSYNC; -const uint32_t C_ATTRIBUTE_LOG = (uint32_t)SCMP_FLTATR_CTL_LOG; -const uint32_t C_ATTRIBUTE_SSB = (uint32_t)SCMP_FLTATR_CTL_SSB; +const uint32_t C_ATTRIBUTE_DEFAULT = (uint32_t)SCMP_FLTATR_ACT_DEFAULT; +const uint32_t C_ATTRIBUTE_BADARCH = (uint32_t)SCMP_FLTATR_ACT_BADARCH; +const uint32_t C_ATTRIBUTE_NNP = (uint32_t)SCMP_FLTATR_CTL_NNP; +const uint32_t C_ATTRIBUTE_TSYNC = (uint32_t)SCMP_FLTATR_CTL_TSYNC; +const uint32_t C_ATTRIBUTE_LOG = (uint32_t)SCMP_FLTATR_CTL_LOG; +const uint32_t C_ATTRIBUTE_SSB = (uint32_t)SCMP_FLTATR_CTL_SSB; +const uint32_t C_ATTRIBUTE_OPTIMIZE = (uint32_t)SCMP_FLTATR_CTL_OPTIMIZE; +const uint32_t C_ATTRIBUTE_SYSRAWRC = (uint32_t)SCMP_FLTATR_API_SYSRAWRC; const int C_CMP_NE = (int)SCMP_CMP_NE; const int C_CMP_LT = (int)SCMP_CMP_LT; @@ -173,8 +182,7 @@ unsigned int get_micro_version() #endif // The libseccomp API level functions were added in v2.4.0 -#if (SCMP_VER_MAJOR < 2) || \ - (SCMP_VER_MAJOR == 2 && SCMP_VER_MINOR < 4) +#if SCMP_VER_MAJOR == 2 && SCMP_VER_MINOR < 4 const unsigned int seccomp_api_get(void) { // libseccomp-golang requires libseccomp v2.2.0, at a minimum, which @@ -217,8 +225,7 @@ void add_struct_arg_cmp( } // The seccomp notify API functions were added in v2.5.0 -#if (SCMP_VER_MAJOR < 2) || \ - (SCMP_VER_MAJOR == 2 && SCMP_VER_MINOR < 5) +#if SCMP_VER_MAJOR == 2 && SCMP_VER_MINOR < 5 struct seccomp_data { int nr; @@ -270,11 +277,13 @@ type scmpFilterAttr uint32 const ( filterAttrActDefault scmpFilterAttr = iota - filterAttrActBadArch scmpFilterAttr = iota - filterAttrNNP scmpFilterAttr = iota - filterAttrTsync scmpFilterAttr = iota - filterAttrLog scmpFilterAttr = iota - filterAttrSSB scmpFilterAttr = iota + filterAttrActBadArch + filterAttrNNP + filterAttrTsync + filterAttrLog + filterAttrSSB + filterAttrOptimize + filterAttrRawRC ) const ( @@ -282,9 +291,9 @@ const ( scmpError C.int = -1 // Comparison boundaries to check for architecture validity archStart ScmpArch = ArchNative - archEnd ScmpArch = ArchPARISC64 + archEnd ScmpArch = ArchRISCV64 // Comparison boundaries to check for action validity - actionStart ScmpAction = ActKill + actionStart ScmpAction = ActKillThread actionEnd ScmpAction = ActKillProcess // Comparison boundaries to check for comparison operator validity compareOpStart ScmpCompareOp = CompareNotEqual @@ -292,8 +301,9 @@ const ( ) var ( - // Error thrown on bad filter context - errBadFilter = fmt.Errorf("filter is invalid or uninitialized") + // errBadFilter is thrown on bad filter context. + errBadFilter = errors.New("filter is invalid or uninitialized") + errDefAction = errors.New("requested action matches default action of filter") // Constants representing library major, minor, and micro versions verMajor = uint(C.get_major_version()) verMinor = uint(C.get_minor_version()) @@ -302,19 +312,28 @@ var ( // Nonexported functions -// Check if library version is greater than or equal to the given one -func checkVersionAbove(major, minor, micro uint) bool { - return (verMajor > major) || +// checkVersion returns an error if the libseccomp version being used +// is less than the one specified by major, minor, and micro arguments. +// Argument op is an arbitrary non-empty operation description, which +// is used as a part of the error message returned. +// +// Most users should use checkAPI instead. +func checkVersion(op string, major, minor, micro uint) error { + if (verMajor > major) || (verMajor == major && verMinor > minor) || - (verMajor == major && verMinor == minor && verMicro >= micro) + (verMajor == major && verMinor == minor && verMicro >= micro) { + return nil + } + return &VersionError{ + op: op, + major: major, + minor: minor, + micro: micro, + } } -// Ensure that the library is supported, i.e. >= 2.2.0. func ensureSupportedVersion() error { - if !checkVersionAbove(2, 2, 0) { - return VersionError{} - } - return nil + return checkVersion("seccomp", 2, 3, 1) } // Get the API level @@ -406,8 +425,10 @@ func (f *ScmpFilter) addRuleWrapper(call ScmpSyscall, action ScmpAction, exact b switch e := errRc(retCode); e { case syscall.EFAULT: return fmt.Errorf("unrecognized syscall %#x", int32(call)) - case syscall.EPERM: - return fmt.Errorf("requested action matches default action of filter") + // libseccomp >= v2.5.0 returns EACCES, older versions return EPERM. + // TODO: remove EPERM once libseccomp < v2.5.0 is not supported. + case syscall.EPERM, syscall.EACCES: + return errDefAction case syscall.EINVAL: return fmt.Errorf("two checks on same syscall argument") default: @@ -432,14 +453,6 @@ func (f *ScmpFilter) addRuleGeneric(call ScmpSyscall, action ScmpAction, exact b return err } } else { - // We don't support conditional filtering in library version v2.1 - if !checkVersionAbove(2, 2, 1) { - return VersionError{ - message: "conditional filtering is not supported", - minimum: "2.2.1", - } - } - argsArr := C.make_arg_cmp_array(C.uint(len(conds))) if argsArr == nil { return fmt.Errorf("error allocating memory for conditions") @@ -536,6 +549,8 @@ func archFromNative(a C.uint32_t) (ScmpArch, error) { return ArchPARISC, nil case C.C_ARCH_PARISC64: return ArchPARISC64, nil + case C.C_ARCH_RISCV64: + return ArchRISCV64, nil default: return 0x0, fmt.Errorf("unrecognized architecture %#x", uint32(a)) } @@ -580,6 +595,8 @@ func (a ScmpArch) toNative() C.uint32_t { return C.C_ARCH_PARISC case ArchPARISC64: return C.C_ARCH_PARISC64 + case ArchRISCV64: + return C.C_ARCH_RISCV64 case ArchNative: return C.C_ARCH_NATIVE default: @@ -612,8 +629,6 @@ func (a ScmpCompareOp) toNative() C.int { func actionFromNative(a C.uint32_t) (ScmpAction, error) { aTmp := a & 0xFFFF switch a & 0xFFFF0000 { - case C.C_ACT_KILL: - return ActKill, nil case C.C_ACT_KILL_PROCESS: return ActKillProcess, nil case C.C_ACT_KILL_THREAD: @@ -638,8 +653,6 @@ func actionFromNative(a C.uint32_t) (ScmpAction, error) { // Only use with sanitized actions, no error handling func (a ScmpAction) toNative() C.uint32_t { switch a & 0xFFFF { - case ActKill: - return C.C_ACT_KILL case ActKillProcess: return C.C_ACT_KILL_PROCESS case ActKillThread: @@ -676,15 +689,15 @@ func (a scmpFilterAttr) toNative() uint32 { return uint32(C.C_ATTRIBUTE_LOG) case filterAttrSSB: return uint32(C.C_ATTRIBUTE_SSB) + case filterAttrOptimize: + return uint32(C.C_ATTRIBUTE_OPTIMIZE) + case filterAttrRawRC: + return uint32(C.C_ATTRIBUTE_SYSRAWRC) default: return 0x0 } } -func (a ScmpSyscall) toNative() C.uint32_t { - return C.uint32_t(a) -} - func syscallFromNative(a C.int) ScmpSyscall { return ScmpSyscall(a) } @@ -724,9 +737,34 @@ func (scmpResp *ScmpNotifResp) toNative(resp *C.struct_seccomp_notif_resp) { resp.flags = C.__u32(scmpResp.Flags) } +// checkAPI checks that both the API level and the seccomp version is equal to +// or greater than the specified minLevel and major, minor, micro, +// respectively, and returns an error otherwise. Argument op is an arbitrary +// non-empty operation description, used as a part of the error message +// returned. +func checkAPI(op string, minLevel uint, major, minor, micro uint) error { + // Ignore error from getAPI, as it returns level == 0 in case of error. + level, _ := getAPI() + if level >= minLevel { + return checkVersion(op, major, minor, micro) + } + return &VersionError{ + op: op, + curAPI: level, + minAPI: minLevel, + major: major, + minor: minor, + micro: micro, + } +} + // Userspace Notification API // Calls to C.seccomp_notify* hidden from seccomp.go +func notifSupported() error { + return checkAPI("seccomp notification", 6, 2, 5, 0) +} + func (f *ScmpFilter) getNotifFd() (ScmpFd, error) { f.lock.Lock() defer f.lock.Unlock() @@ -734,11 +772,8 @@ func (f *ScmpFilter) getNotifFd() (ScmpFd, error) { if !f.valid { return -1, errBadFilter } - - // Ignore error, if not supported returns apiLevel == 0 - apiLevel, _ := GetAPI() - if apiLevel < 6 { - return -1, fmt.Errorf("seccomp notification requires API level >= 6; current level = %d", apiLevel) + if err := notifSupported(); err != nil { + return -1, err } fd := C.seccomp_notify_fd(f.filterCtx) @@ -750,10 +785,8 @@ func notifReceive(fd ScmpFd) (*ScmpNotifReq, error) { var req *C.struct_seccomp_notif var resp *C.struct_seccomp_notif_resp - // Ignore error, if not supported returns apiLevel == 0 - apiLevel, _ := GetAPI() - if apiLevel < 6 { - return nil, fmt.Errorf("seccomp notification requires API level >= 6; current level = %d", apiLevel) + if err := notifSupported(); err != nil { + return nil, err } // we only use the request here; the response is unused @@ -789,13 +822,11 @@ func notifRespond(fd ScmpFd, scmpResp *ScmpNotifResp) error { var req *C.struct_seccomp_notif var resp *C.struct_seccomp_notif_resp - // Ignore error, if not supported returns apiLevel == 0 - apiLevel, _ := GetAPI() - if apiLevel < 6 { - return fmt.Errorf("seccomp notification requires API level >= 6; current level = %d", apiLevel) + if err := notifSupported(); err != nil { + return err } - // we only use the reponse here; the request is discarded + // we only use the response here; the request is discarded if retCode := C.seccomp_notify_alloc(&req, &resp); retCode != 0 { return errRc(retCode) } @@ -827,10 +858,8 @@ func notifRespond(fd ScmpFd, scmpResp *ScmpNotifResp) error { } func notifIDValid(fd ScmpFd, id uint64) error { - // Ignore error, if not supported returns apiLevel == 0 - apiLevel, _ := GetAPI() - if apiLevel < 6 { - return fmt.Errorf("seccomp notification requires API level >= 6; current level = %d", apiLevel) + if err := notifSupported(); err != nil { + return err } for { diff --git a/vendor/modules.txt b/vendor/modules.txt index d21045c0d..de76aad91 100644 --- a/vendor/modules.txt +++ b/vendor/modules.txt @@ -555,7 +555,7 @@ github.com/opencontainers/go-digest ## explicit github.com/opencontainers/image-spec/specs-go github.com/opencontainers/image-spec/specs-go/v1 -# github.com/opencontainers/runc v1.1.2 +# github.com/opencontainers/runc v1.1.3 ## explicit github.com/opencontainers/runc/libcontainer/apparmor github.com/opencontainers/runc/libcontainer/cgroups @@ -628,7 +628,7 @@ github.com/rootless-containers/rootlesskit/pkg/port/builtin/parent/tcp github.com/rootless-containers/rootlesskit/pkg/port/builtin/parent/udp github.com/rootless-containers/rootlesskit/pkg/port/builtin/parent/udp/udpproxy github.com/rootless-containers/rootlesskit/pkg/port/portutil -# github.com/seccomp/libseccomp-golang v0.9.2-0.20210429002308-3879420cc921 +# github.com/seccomp/libseccomp-golang v0.9.2-0.20220502022130-f33da4d89646 github.com/seccomp/libseccomp-golang # github.com/sirupsen/logrus v1.8.1 ## explicit |