Merge pull request #9536 from jmguzik/enable-cgroupsv2-sec-opts

Enable cgroupsv2 rw mount via security-opt unmask

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

@@ -859,7 +859,7 @@ Security Options
 
 - `unmask=ALL or /path/1:/path/2` : Paths to unmask separated by a colon. If set to **ALL**, it will
   unmask all the paths that are masked or made read only by default.
-  The default masked paths are **/proc/acpi, /proc/kcore, /proc/keys, /proc/latency_stats, /proc/sched_debug, /proc/scsi, /proc/timer_list, /proc/timer_stats, /sys/firmware, and /sys/fs/selinux.**  The default paths that are read only are **/proc/asound, /proc/bus, /proc/fs, /proc/irq, /proc/sys, /proc/sysrq-trigger**.
+  The default masked paths are **/proc/acpi, /proc/kcore, /proc/keys, /proc/latency_stats, /proc/sched_debug, /proc/scsi, /proc/timer_list, /proc/timer_stats, /sys/firmware, and /sys/fs/selinux.**  The default paths that are read only are **/proc/asound, /proc/bus, /proc/fs, /proc/irq, /proc/sys, /proc/sysrq-trigger, /sys/fs/cgroup**.
 
 - `proc-opts=OPTIONS` : Comma separated list of options to use for the /proc mount. More details for the
   possible mount options are specified at **proc(5)** man page.

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

@@ -908,7 +908,7 @@ Security Options
   for the possible mount options are specified at **proc(5)** man page.
 - **unmask**=_ALL_ or _/path/1:/path/2_: Paths to unmask separated by a colon. If set to **ALL**, it will
   unmask all the paths that are masked or made read only by default.
-  The default masked paths are **/proc/acpi, /proc/kcore, /proc/keys, /proc/latency_stats, /proc/sched_debug, /proc/scsi, /proc/timer_list, /proc/timer_stats, /sys/firmware, and /sys/fs/selinux.**.  The default paths that are read only are **/proc/asound**, **/proc/bus**, **/proc/fs**, **/proc/irq**, **/proc/sys**, **/proc/sysrq-trigger**.
+  The default masked paths are **/proc/acpi, /proc/kcore, /proc/keys, /proc/latency_stats, /proc/sched_debug, /proc/scsi, /proc/timer_list, /proc/timer_stats, /sys/firmware, and /sys/fs/selinux.**.  The default paths that are read only are **/proc/asound**, **/proc/bus**, **/proc/fs**, **/proc/irq**, **/proc/sys**, **/proc/sysrq-trigger**, **/sys/fs/cgroup**.
 
 Note: Labeling can be disabled for all containers by setting **label=false** in the **containers.conf**(5) file.
 

pkg/specgen/generate/oci.go

@@ -2,12 +2,14 @@ package generate
 
 import (
 	"context"
+	"path"
 	"strings"
 
 	"github.com/containers/common/pkg/config"
 	"github.com/containers/podman/v3/libpod"
 	"github.com/containers/podman/v3/libpod/define"
 	"github.com/containers/podman/v3/libpod/image"
+	"github.com/containers/podman/v3/pkg/cgroups"
 	"github.com/containers/podman/v3/pkg/rootless"
 	"github.com/containers/podman/v3/pkg/specgen"
 	spec "github.com/opencontainers/runtime-spec/specs-go"
@@ -157,8 +159,32 @@ func canMountSys(isRootless, isNewUserns bool, s *specgen.SpecGenerator) bool {
 	return true
 }
 
+func getCGroupPermissons(unmask []string) string {
+	ro := "ro"
+	rw := "rw"
+	cgroup := "/sys/fs/cgroup"
+
+	cgroupv2, _ := cgroups.IsCgroup2UnifiedMode()
+	if !cgroupv2 {
+		return ro
+	}
+
+	if unmask != nil && unmask[0] == "ALL" {
+		return rw
+	}
+
+	for _, p := range unmask {
+		if path.Clean(p) == cgroup {
+			return rw
+		}
+	}
+	return ro
+}
+
+// SpecGenToOCI returns the base configuration for the container.
 func SpecGenToOCI(ctx context.Context, s *specgen.SpecGenerator, rt *libpod.Runtime, rtc *config.Config, newImage *image.Image, mounts []spec.Mount, pod *libpod.Pod, finalCmd []string) (*spec.Spec, error) {
-	cgroupPerm := "ro"
+	cgroupPerm := getCGroupPermissons(s.Unmask)
+
 	g, err := generate.New("linux")
 	if err != nil {
 		return nil, err

test/e2e/run_test.go

@@ -304,6 +304,42 @@ var _ = Describe("Podman run", func() {
 
 	})
 
+	It("podman run security-opt unmask on /sys/fs/cgroup", func() {
+
+		SkipIfCgroupV1("podman umask on /sys/fs/cgroup will fail with cgroups V1")
+		SkipIfRootless("/sys/fs/cgroup rw access is needed")
+		rwOnCGroups := "/sys/fs/cgroup cgroup2 rw"
+		session := podmanTest.Podman([]string{"run", "--security-opt", "unmask=ALL", "--security-opt", "mask=/sys/fs/cgroup", ALPINE, "cat", "/proc/mounts"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+		Expect(session.OutputToString()).To(ContainSubstring(rwOnCGroups))
+
+		session = podmanTest.Podman([]string{"run", "--security-opt", "unmask=/sys/fs/cgroup", ALPINE, "cat", "/proc/mounts"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+		Expect(session.OutputToString()).To(ContainSubstring(rwOnCGroups))
+
+		session = podmanTest.Podman([]string{"run", "--security-opt", "unmask=/sys/fs/cgroup///", ALPINE, "cat", "/proc/mounts"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+		Expect(session.OutputToString()).To(ContainSubstring(rwOnCGroups))
+
+		session = podmanTest.Podman([]string{"run", "--security-opt", "unmask=ALL", ALPINE, "cat", "/proc/mounts"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+		Expect(session.OutputToString()).To(ContainSubstring(rwOnCGroups))
+
+		session = podmanTest.Podman([]string{"run", "--security-opt", "unmask=/sys/fs/cgroup", "--security-opt", "mask=/sys/fs/cgroup", ALPINE, "cat", "/proc/mounts"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+		Expect(session.OutputToString()).To(ContainSubstring(rwOnCGroups))
+
+		session = podmanTest.Podman([]string{"run", "--security-opt", "unmask=/sys/fs/cgroup", ALPINE, "ls", "/sys/fs/cgroup"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+		Expect(session.OutputToString()).ToNot(BeEmpty())
+	})
+
 	It("podman run seccomp test", func() {
 		session := podmanTest.Podman([]string{"run", "-it", "--security-opt", strings.Join([]string{"seccomp=", forbidGetCWDSeccompProfile()}, ""), ALPINE, "pwd"})
 		session.WaitWithDefaultTimeout()