Merge pull request #7372 from giuseppe/add-unified-configuration

podman: add option --cgroup-conf

cmd/podman/common/create.go

@@ -516,5 +516,10 @@ func GetCreateFlags(cf *ContainerCLIOpts) *pflag.FlagSet {
 		"seccomp-policy", "default",
 		"Policy for selecting a seccomp profile (experimental)",
 	)
+	createFlags.StringSliceVar(
+		&cf.CgroupConf,
+		"cgroup-conf", []string{},
+		"Configure cgroup v2 (key=value)",
+	)
 	return &createFlags
 }

cmd/podman/common/create_opts.go

@@ -106,4 +106,6 @@ type ContainerCLIOpts struct {
 	SeccompPolicy     string
 
 	Net *entities.NetOptions
+
+	CgroupConf []string
 }

cmd/podman/common/specgen.go

@@ -450,7 +450,20 @@ func FillOutSpecGen(s *specgen.SpecGenerator, c *ContainerCLIOpts, args []string
 		s.ResourceLimits.Pids = &pids
 	}
 	s.ResourceLimits.CPU = getCPULimits(c)
-	if s.ResourceLimits.CPU == nil && s.ResourceLimits.Pids == nil && s.ResourceLimits.BlockIO == nil && s.ResourceLimits.Memory == nil {
+
+	unifieds := make(map[string]string)
+	for _, unified := range c.CgroupConf {
+		splitUnified := strings.SplitN(unified, "=", 2)
+		if len(splitUnified) < 2 {
+			return errors.Errorf("--cgroup-conf must be formatted KEY=VALUE")
+		}
+		unifieds[splitUnified[0]] = splitUnified[1]
+	}
+	if len(unifieds) > 0 {
+		s.ResourceLimits.Unified = unifieds
+	}
+
+	if s.ResourceLimits.CPU == nil && s.ResourceLimits.Pids == nil && s.ResourceLimits.BlockIO == nil && s.ResourceLimits.Memory == nil && s.ResourceLimits.Unified == nil {
 		s.ResourceLimits = nil
 	}
 

completions/bash/podman

@@ -2107,6 +2107,7 @@ _podman_container_run() {
 		--cap-add
 		--cap-drop
 		--cgroup-parent
+		--cgroup-conf
 		--cidfile
 		--conmon-pidfile
 		--cpu-period

docs/source/markdown/podman-create.1.md

@@ -89,6 +89,10 @@ The *split* option splits the current cgroup in two sub-cgroups: one for conmon
 
 Path to cgroups under which the cgroup for the container will be created. If the path is not absolute, the path is considered to be relative to the cgroups path of the init process. Cgroups will be created if they do not already exist.
 
+**--cgroup-conf**=*KEY=VALUE*
+
+When running on cgroup v2, specify the cgroup file to write to and its value.  For example **--cgroup-conf=memory.high=1073741824** sets the memory.high limit to 1GB.
+
 **--cidfile**=*id*
 
 Write the container ID to the file

docs/source/markdown/podman-run.1.md

@@ -104,6 +104,10 @@ The **split** option splits the current cgroup in two sub-cgroups: one for conmo
 
 Path to cgroups under which the cgroup for the container will be created. If the path is not absolute, the path is considered to be relative to the cgroups path of the init process. Cgroups will be created if they do not already exist.
 
+**--cgroup-conf**=*KEY=VALUE*
+
+When running on cgroup v2, specify the cgroup file to write to and its value.  For example **--cgroup-conf=memory.high=1073741824** sets the memory.high limit to 1GB.
+
 **--cidfile**=*file*
 
 Write the container ID to *file*.

go.mod

@@ -40,8 +40,8 @@ require (
 	github.com/opencontainers/go-digest v1.0.0
 	github.com/opencontainers/image-spec v1.0.2-0.20190823105129-775207bd45b6
 	github.com/opencontainers/runc v1.0.0-rc91.0.20200708210054-ce54a9d4d79b
-	github.com/opencontainers/runtime-spec v1.0.3-0.20200520003142-237cc4f519e2
-	github.com/opencontainers/runtime-tools v0.9.1-0.20200714183735-07406c5828aa
+	github.com/opencontainers/runtime-spec v1.0.3-0.20200817204227-f9c09b4ea1df
+	github.com/opencontainers/runtime-tools v0.9.0
 	github.com/opencontainers/selinux v1.6.0
 	github.com/opentracing/opentracing-go v1.2.0
 	github.com/pkg/errors v0.9.1

go.sum

@@ -342,10 +342,11 @@ github.com/opencontainers/runc v1.0.0-rc91.0.20200708210054-ce54a9d4d79b/go.mod
 github.com/opencontainers/runtime-spec v0.1.2-0.20190507144316-5b71a03e2700/go.mod h1:jwyrGlmzljRJv/Fgzds9SsS/C5hL+LL3ko9hs6T5lQ0=
 github.com/opencontainers/runtime-spec v1.0.3-0.20200520003142-237cc4f519e2 h1:9mv9SC7GWmRWE0J/+oD8w3GsN2KYGKtg6uwLN7hfP5E=
 github.com/opencontainers/runtime-spec v1.0.3-0.20200520003142-237cc4f519e2/go.mod h1:jwyrGlmzljRJv/Fgzds9SsS/C5hL+LL3ko9hs6T5lQ0=
+github.com/opencontainers/runtime-spec v1.0.3-0.20200817204227-f9c09b4ea1df h1:5AW5dMFSXVH4Mg3WYe4z7ui64bK8n66IoWK8i6T4QZ8=
+github.com/opencontainers/runtime-spec v1.0.3-0.20200817204227-f9c09b4ea1df/go.mod h1:jwyrGlmzljRJv/Fgzds9SsS/C5hL+LL3ko9hs6T5lQ0=
 github.com/opencontainers/runtime-tools v0.0.0-20181011054405-1d69bd0f9c39/go.mod h1:r3f7wjNzSs2extwzU3Y+6pKfobzPh+kKFJ3ofN+3nfs=
+github.com/opencontainers/runtime-tools v0.9.0 h1:FYgwVsKRI/H9hU32MJ/4MLOzXWodKK5zsQavY8NPMkU=
 github.com/opencontainers/runtime-tools v0.9.0/go.mod h1:r3f7wjNzSs2extwzU3Y+6pKfobzPh+kKFJ3ofN+3nfs=
-github.com/opencontainers/runtime-tools v0.9.1-0.20200714183735-07406c5828aa h1:iyj+fFHVBn0xOalz9UChYzSU1K0HJ+d75b4YqShBRhI=
-github.com/opencontainers/runtime-tools v0.9.1-0.20200714183735-07406c5828aa/go.mod h1:r3f7wjNzSs2extwzU3Y+6pKfobzPh+kKFJ3ofN+3nfs=
 github.com/opencontainers/selinux v1.3.0/go.mod h1:+BLncwf63G4dgOzykXAxcmnFlUaOlkDdmw/CqsW6pjs=
 github.com/opencontainers/selinux v1.5.1/go.mod h1:yTcKuYAh6R95iDpefGLQaPaRwJFwyzAJufJyiTt7s0g=
 github.com/opencontainers/selinux v1.5.2/go.mod h1:yTcKuYAh6R95iDpefGLQaPaRwJFwyzAJufJyiTt7s0g=

libpod/container_inspect.go

@@ -465,6 +465,7 @@ func (c *Container) generateInspectContainerHostConfig(ctrSpec *spec.Spec, named
 			if ctrSpec.Linux.Resources.Pids != nil {
 				hostConfig.PidsLimit = ctrSpec.Linux.Resources.Pids.Limit
 			}
+			hostConfig.CgroupConf = ctrSpec.Linux.Resources.Unified
 			if ctrSpec.Linux.Resources.BlockIO != nil {
 				if ctrSpec.Linux.Resources.BlockIO.Weight != nil {
 					hostConfig.BlkioWeight = *ctrSpec.Linux.Resources.BlockIO.Weight

libpod/container_internal_linux.go

@@ -385,7 +385,8 @@ func (c *Container) generateSpec(ctx context.Context) (*spec.Spec, error) {
 		if err != nil {
 			return nil, errors.Wrapf(err, "Invalid Umask Value")
 		}
-		g.SetProcessUmask(uint32(decVal))
+		umask := uint32(decVal)
+		g.Config.Process.User.Umask = &umask
 	}
 
 	// Add addition groups if c.config.GroupAdd is not empty

libpod/define/container_inspect.go

@@ -518,6 +518,8 @@ type InspectContainerHostConfig struct {
 	IOMaximumIOps uint64 `json:"IOMaximumIOps"`
 	// IOMaximumBandwidth is Windows-only and not presently implemented.
 	IOMaximumBandwidth uint64 `json:"IOMaximumBandwidth"`
+	// CgroupConf is the configuration for cgroup v2.
+	CgroupConf map[string]string `json:"CgroupConf"`
 }
 
 // InspectBasicNetworkConfig holds basic configuration information (e.g. IP

pkg/spec/createconfig.go

@@ -31,12 +31,13 @@ const (
 type CreateResourceConfig struct {
 	BlkioWeight       uint16   // blkio-weight
 	BlkioWeightDevice []string // blkio-weight-device
-	CPUPeriod         uint64   // cpu-period
-	CPUQuota          int64    // cpu-quota
-	CPURtPeriod       uint64   // cpu-rt-period
-	CPURtRuntime      int64    // cpu-rt-runtime
-	CPUShares         uint64   // cpu-shares
-	CPUs              float64  // cpus
+	CgroupConf        map[string]string
+	CPUPeriod         uint64  // cpu-period
+	CPUQuota          int64   // cpu-quota
+	CPURtPeriod       uint64  // cpu-rt-period
+	CPURtRuntime      int64   // cpu-rt-runtime
+	CPUShares         uint64  // cpu-shares
+	CPUs              float64 // cpus
 	CPUsetCPUs        string
 	CPUsetMems        string   // cpuset-mems
 	DeviceCgroupRules []string //device-cgroup-rule

pkg/specgen/generate/validate.go

@@ -23,6 +23,12 @@ func verifyContainerResources(s *specgen.SpecGenerator) ([]string, error) {
 		return warnings, nil
 	}
 
+	if s.ResourceLimits.Unified != nil {
+		if !cgroup2 {
+			return nil, errors.New("Cannot use --cgroup-conf without cgroup v2")
+		}
+	}
+
 	// Memory checks
 	if s.ResourceLimits.Memory != nil {
 		memory := s.ResourceLimits.Memory

pkg/specgen/specgen.go

@@ -415,6 +415,10 @@ type ContainerResourceConfig struct {
 	ThrottleReadIOPSDevice map[string]spec.LinuxThrottleDevice `json:"throttleReadIOPSDevice,omitempty"`
 	// IO write rate limit per cgroup per device, IO per second
 	ThrottleWriteIOPSDevice map[string]spec.LinuxThrottleDevice `json:"throttleWriteIOPSDevice,omitempty"`
+	// CgroupConf are key-value options passed into the container runtime
+	// that are used to configure cgroup v2.
+	// Optional.
+	CgroupConf map[string]string `json:"unified,omitempty"`
 }
 
 // ContainerHealthCheckConfig describes a container healthcheck with attributes

vendor/github.com/opencontainers/runtime-spec/specs-go/config.go

@@ -60,7 +60,7 @@ type Process struct {
 	SelinuxLabel string `json:"selinuxLabel,omitempty" platform:"linux"`
 }
 
-// LinuxCapabilities specifies the whitelist of capabilities that are kept for a process.
+// LinuxCapabilities specifies the list of allowed capabilities that are kept for a process.
 // http://man7.org/linux/man-pages/man7/capabilities.7.html
 type LinuxCapabilities struct {
 	// Bounding is the set of capabilities checked by the kernel.
@@ -90,7 +90,7 @@ type User struct {
 	// GID is the group id.
 	GID uint32 `json:"gid" platform:"linux,solaris"`
 	// Umask is the umask for the init process.
-	Umask uint32 `json:"umask,omitempty" platform:"linux,solaris"`
+	Umask *uint32 `json:"umask,omitempty" platform:"linux,solaris"`
 	// AdditionalGids are additional group ids set for the container's process.
 	AdditionalGids []uint32 `json:"additionalGids,omitempty" platform:"linux,solaris"`
 	// Username is the user name.
@@ -354,7 +354,7 @@ type LinuxRdma struct {
 
 // LinuxResources has container runtime resource constraints
 type LinuxResources struct {
-	// Devices configures the device whitelist.
+	// Devices configures the device allowlist.
 	Devices []LinuxDeviceCgroup `json:"devices,omitempty"`
 	// Memory restriction configuration
 	Memory *LinuxMemory `json:"memory,omitempty"`
@@ -372,6 +372,8 @@ type LinuxResources struct {
 	// Limits are a set of key value pairs that define RDMA resource limits,
 	// where the key is device name and value is resource limits.
 	Rdma map[string]LinuxRdma `json:"rdma,omitempty"`
+	// Unified resources.
+	Unified map[string]string `json:"unified,omitempty"`
 }
 
 // LinuxDevice represents the mknod information for a Linux special device file
@@ -392,7 +394,8 @@ type LinuxDevice struct {
 	GID *uint32 `json:"gid,omitempty"`
 }
 
-// LinuxDeviceCgroup represents a device rule for the whitelist controller
+// LinuxDeviceCgroup represents a device rule for the devices specified to
+// the device controller
 type LinuxDeviceCgroup struct {
 	// Allow or deny
 	Allow bool `json:"allow"`
@@ -628,6 +631,7 @@ const (
 	ArchS390X       Arch = "SCMP_ARCH_S390X"
 	ArchPARISC      Arch = "SCMP_ARCH_PARISC"
 	ArchPARISC64    Arch = "SCMP_ARCH_PARISC64"
+	ArchRISCV64     Arch = "SCMP_ARCH_RISCV64"
 )
 
 // LinuxSeccompAction taken upon Seccomp rule match
@@ -635,12 +639,13 @@ type LinuxSeccompAction string
 
 // Define actions for Seccomp rules
 const (
-	ActKill  LinuxSeccompAction = "SCMP_ACT_KILL"
-	ActTrap  LinuxSeccompAction = "SCMP_ACT_TRAP"
-	ActErrno LinuxSeccompAction = "SCMP_ACT_ERRNO"
-	ActTrace LinuxSeccompAction = "SCMP_ACT_TRACE"
-	ActAllow LinuxSeccompAction = "SCMP_ACT_ALLOW"
-	ActLog   LinuxSeccompAction = "SCMP_ACT_LOG"
+	ActKill        LinuxSeccompAction = "SCMP_ACT_KILL"
+	ActKillProcess LinuxSeccompAction = "SCMP_ACT_KILL_PROCESS"
+	ActTrap        LinuxSeccompAction = "SCMP_ACT_TRAP"
+	ActErrno       LinuxSeccompAction = "SCMP_ACT_ERRNO"
+	ActTrace       LinuxSeccompAction = "SCMP_ACT_TRACE"
+	ActAllow       LinuxSeccompAction = "SCMP_ACT_ALLOW"
+	ActLog         LinuxSeccompAction = "SCMP_ACT_LOG"
 )
 
 // LinuxSeccompOperator used to match syscall arguments in Seccomp

vendor/github.com/opencontainers/runtime-spec/specs-go/state.go

@@ -1,5 +1,23 @@
 package specs
 
+// ContainerState represents the state of a container.
+type ContainerState string
+
+const (
+	// StateCreating indicates that the container is being created
+	StateCreating ContainerState  = "creating"
+
+	// StateCreated indicates that the runtime has finished the create operation
+	StateCreated ContainerState  = "created"
+
+	// StateRunning indicates that the container process has executed the
+	// user-specified program but has not exited
+	StateRunning ContainerState  = "running"
+
+	// StateStopped indicates that the container process has exited
+	StateStopped ContainerState  = "stopped"
+)
+
 // State holds information about the runtime state of the container.
 type State struct {
 	// Version is the version of the specification that is supported.
@@ -7,7 +25,7 @@ type State struct {
 	// ID is the container ID
 	ID string `json:"id"`
 	// Status is the runtime status of the container.
-	Status string `json:"status"`
+	Status ContainerState `json:"status"`
 	// Pid is the process ID for the container process.
 	Pid int `json:"pid,omitempty"`
 	// Bundle is the path to the container's bundle directory.

vendor/github.com/opencontainers/runtime-tools/generate/generate.go

@@ -29,9 +29,6 @@ var (
 type Generator struct {
 	Config       *rspec.Spec
 	HostSpecific bool
-	// This is used to keep a cache of the ENVs added to improve
-	// performance when adding a huge number of ENV variables
-	envMap map[string]int
 }
 
 // ExportOptions have toggles for exporting only certain parts of the specification
@@ -239,12 +236,7 @@ func New(os string) (generator Generator, err error) {
 		}
 	}
 
-	envCache := map[string]int{}
-	if config.Process != nil {
-		envCache = createEnvCacheMap(config.Process.Env)
-	}
-
-	return Generator{Config: &config, envMap: envCache}, nil
+	return Generator{Config: &config}, nil
 }
 
 // NewFromSpec creates a configuration Generator from a given
@@ -254,14 +246,8 @@ func New(os string) (generator Generator, err error) {
 //
 //   generator := Generator{Config: config}
 func NewFromSpec(config *rspec.Spec) Generator {
-	envCache := map[string]int{}
-	if config != nil && config.Process != nil {
-		envCache = createEnvCacheMap(config.Process.Env)
-	}
-
 	return Generator{
 		Config: config,
-		envMap: envCache,
 	}
 }
 
@@ -287,27 +273,11 @@ func NewFromTemplate(r io.Reader) (Generator, error) {
 	if err := json.NewDecoder(r).Decode(&config); err != nil {
 		return Generator{}, err
 	}
-
-	envCache := map[string]int{}
-	if config.Process != nil {
-		envCache = createEnvCacheMap(config.Process.Env)
-	}
-
 	return Generator{
 		Config: &config,
-		envMap: envCache,
 	}, nil
 }
 
-// createEnvCacheMap creates a hash map with the ENV variables given by the config
-func createEnvCacheMap(env []string) map[string]int {
-	envMap := make(map[string]int, len(env))
-	for i, val := range env {
-		envMap[val] = i
-	}
-	return envMap
-}
-
 // SetSpec sets the configuration in the Generator g.
 //
 // Deprecated: Replace with:
@@ -444,12 +414,6 @@ func (g *Generator) SetProcessUsername(username string) {
 	g.Config.Process.User.Username = username
 }
 
-// SetProcessUmask sets g.Config.Process.User.Umask.
-func (g *Generator) SetProcessUmask(umask uint32) {
-	g.initConfigProcess()
-	g.Config.Process.User.Umask = umask
-}
-
 // SetProcessGID sets g.Config.Process.User.GID.
 func (g *Generator) SetProcessGID(gid uint32) {
 	g.initConfigProcess()
@@ -492,44 +456,21 @@ func (g *Generator) ClearProcessEnv() {
 		return
 	}
 	g.Config.Process.Env = []string{}
-	// Clear out the env cache map as well
-	g.envMap = map[string]int{}
 }
 
 // AddProcessEnv adds name=value into g.Config.Process.Env, or replaces an
 // existing entry with the given name.
 func (g *Generator) AddProcessEnv(name, value string) {
-	if name == "" {
-		return
-	}
-
-	g.initConfigProcess()
-	g.addEnv(fmt.Sprintf("%s=%s", name, value), name)
-}
-
-// AddMultipleProcessEnv adds multiple name=value into g.Config.Process.Env, or replaces
-// existing entries with the given name.
-func (g *Generator) AddMultipleProcessEnv(envs []string) {
 	g.initConfigProcess()
 
-	for _, val := range envs {
-		split := strings.SplitN(val, "=", 2)
-		g.addEnv(val, split[0])
-	}
-}
-
-// addEnv looks through adds ENV to the Process and checks envMap for
-// any duplicates
-// This is called by both AddMultipleProcessEnv and AddProcessEnv
-func (g *Generator) addEnv(env, key string) {
-	if idx, ok := g.envMap[key]; ok {
-		// The ENV exists in the cache, so change its value in g.Config.Process.Env
-		g.Config.Process.Env[idx] = env
-	} else {
-		// else the env doesn't exist, so add it and add it's index to g.envMap
-		g.Config.Process.Env = append(g.Config.Process.Env, env)
-		g.envMap[key] = len(g.Config.Process.Env) - 1
+	env := fmt.Sprintf("%s=%s", name, value)
+	for idx := range g.Config.Process.Env {
+		if strings.HasPrefix(g.Config.Process.Env[idx], name+"=") {
+			g.Config.Process.Env[idx] = env
+			return
+		}
 	}
+	g.Config.Process.Env = append(g.Config.Process.Env, env)
 }
 
 // AddProcessRlimits adds rlimit into g.Config.Process.Rlimits.
@@ -1502,7 +1443,7 @@ func (g *Generator) AddDevice(device rspec.LinuxDevice) {
 			return
 		}
 		if dev.Type == device.Type && dev.Major == device.Major && dev.Minor == device.Minor {
-			fmt.Fprintf(os.Stderr, "WARNING: Creating device %q with same type, major and minor as existing %q.\n", device.Path, dev.Path)
+			fmt.Fprintln(os.Stderr, "WARNING: The same type, major and minor should not be used for multiple devices.")
 		}
 	}
 

vendor/github.com/opencontainers/runtime-tools/generate/seccomp/seccomp_default.go

@@ -566,20 +566,6 @@ func DefaultProfile(rs *specs.Spec) *rspec.LinuxSeccomp {
 			},
 		}...)
 		/* Flags parameter of the clone syscall is the 2nd on s390 */
-		syscalls = append(syscalls, []rspec.LinuxSyscall{
-			{
-				Names:  []string{"clone"},
-				Action: rspec.ActAllow,
-				Args: []rspec.LinuxSeccompArg{
-					{
-						Index:    1,
-						Value:    2080505856,
-						ValueTwo: 0,
-						Op:       rspec.OpMaskedEqual,
-					},
-				},
-			},
-		}...)
 	}
 
 	return &rspec.LinuxSeccomp{

vendor/modules.txt

@@ -418,9 +418,9 @@ github.com/opencontainers/runc/libcontainer/devices
 github.com/opencontainers/runc/libcontainer/system
 github.com/opencontainers/runc/libcontainer/user
 github.com/opencontainers/runc/libcontainer/utils
-# github.com/opencontainers/runtime-spec v1.0.3-0.20200520003142-237cc4f519e2
+# github.com/opencontainers/runtime-spec v1.0.3-0.20200817204227-f9c09b4ea1df
 github.com/opencontainers/runtime-spec/specs-go
-# github.com/opencontainers/runtime-tools v0.9.1-0.20200714183735-07406c5828aa
+# github.com/opencontainers/runtime-tools v0.9.0
 github.com/opencontainers/runtime-tools/error
 github.com/opencontainers/runtime-tools/filepath
 github.com/opencontainers/runtime-tools/generate