1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
|
// +build linux
package libpod
import (
"fmt"
"os"
"os/exec"
"path/filepath"
"runtime"
"strings"
"syscall"
"github.com/containerd/cgroups"
"github.com/containers/libpod/pkg/rootless"
"github.com/containers/libpod/utils"
pmount "github.com/containers/storage/pkg/mount"
spec "github.com/opencontainers/runtime-spec/specs-go"
"github.com/pkg/errors"
"github.com/sirupsen/logrus"
"golang.org/x/sys/unix"
)
const unknownPackage = "Unknown"
func (r *OCIRuntime) moveConmonToCgroup(ctr *Container, cgroupParent string, cmd *exec.Cmd) error {
if os.Geteuid() == 0 {
if r.cgroupManager == SystemdCgroupsManager {
unitName := createUnitName("libpod-conmon", ctr.ID())
realCgroupParent := cgroupParent
splitParent := strings.Split(cgroupParent, "/")
if strings.HasSuffix(cgroupParent, ".slice") && len(splitParent) > 1 {
realCgroupParent = splitParent[len(splitParent)-1]
}
logrus.Infof("Running conmon under slice %s and unitName %s", realCgroupParent, unitName)
if err := utils.RunUnderSystemdScope(cmd.Process.Pid, realCgroupParent, unitName); err != nil {
logrus.Warnf("Failed to add conmon to systemd sandbox cgroup: %v", err)
}
} else {
cgroupPath := filepath.Join(ctr.config.CgroupParent, "conmon")
control, err := cgroups.New(cgroups.V1, cgroups.StaticPath(cgroupPath), &spec.LinuxResources{})
if err != nil {
logrus.Warnf("Failed to add conmon to cgroupfs sandbox cgroup: %v", err)
} else {
// we need to remove this defer and delete the cgroup once conmon exits
// maybe need a conmon monitor?
if err := control.Add(cgroups.Process{Pid: cmd.Process.Pid}); err != nil {
logrus.Warnf("Failed to add conmon to cgroupfs sandbox cgroup: %v", err)
}
}
}
}
return nil
}
// newPipe creates a unix socket pair for communication
func newPipe() (parent *os.File, child *os.File, err error) {
fds, err := unix.Socketpair(unix.AF_LOCAL, unix.SOCK_STREAM|unix.SOCK_CLOEXEC, 0)
if err != nil {
return nil, nil, err
}
return os.NewFile(uintptr(fds[1]), "parent"), os.NewFile(uintptr(fds[0]), "child"), nil
}
// makeAccessible changes the path permission and each parent directory to have --x--x--x
func makeAccessible(path string, uid, gid int) error {
for ; path != "/"; path = filepath.Dir(path) {
st, err := os.Stat(path)
if err != nil {
if os.IsNotExist(err) {
return nil
}
return err
}
if int(st.Sys().(*syscall.Stat_t).Uid) == uid && int(st.Sys().(*syscall.Stat_t).Gid) == gid {
continue
}
if st.Mode()&0111 != 0111 {
if err := os.Chmod(path, os.FileMode(st.Mode()|0111)); err != nil {
return err
}
}
}
return nil
}
// CreateContainer creates a container in the OCI runtime
// TODO terminal support for container
// Presently just ignoring conmon opts related to it
func (r *OCIRuntime) createContainer(ctr *Container, cgroupParent string, restoreOptions *ContainerCheckpointOptions) (err error) {
if len(ctr.config.IDMappings.UIDMap) != 0 || len(ctr.config.IDMappings.GIDMap) != 0 {
for _, i := range []string{ctr.state.RunDir, ctr.runtime.config.TmpDir, ctr.config.StaticDir, ctr.state.Mountpoint, ctr.runtime.config.VolumePath} {
if err := makeAccessible(i, ctr.RootUID(), ctr.RootGID()); err != nil {
return err
}
}
// if we are running a non privileged container, be sure to umount some kernel paths so they are not
// bind mounted inside the container at all.
if !ctr.config.Privileged && !rootless.IsRootless() {
ch := make(chan error)
go func() {
runtime.LockOSThread()
err := func() error {
fd, err := os.Open(fmt.Sprintf("/proc/%d/task/%d/ns/mnt", os.Getpid(), unix.Gettid()))
if err != nil {
return err
}
defer fd.Close()
// create a new mountns on the current thread
if err = unix.Unshare(unix.CLONE_NEWNS); err != nil {
return err
}
defer unix.Setns(int(fd.Fd()), unix.CLONE_NEWNS)
// don't spread our mounts around. We are setting only /sys to be slave
// so that the cleanup process is still able to umount the storage and the
// changes are propagated to the host.
err = unix.Mount("/sys", "/sys", "none", unix.MS_REC|unix.MS_SLAVE, "")
if err != nil {
return errors.Wrapf(err, "cannot make /sys slave")
}
mounts, err := pmount.GetMounts()
if err != nil {
return err
}
for _, m := range mounts {
if !strings.HasPrefix(m.Mountpoint, "/sys/kernel") {
continue
}
err = unix.Unmount(m.Mountpoint, 0)
if err != nil {
return errors.Wrapf(err, "cannot unmount %s", m.Mountpoint)
}
}
return r.createOCIContainer(ctr, cgroupParent, restoreOptions)
}()
ch <- err
}()
err := <-ch
return err
}
}
return r.createOCIContainer(ctr, cgroupParent, restoreOptions)
}
func rpmVersion(path string) string {
output := unknownPackage
cmd := exec.Command("/usr/bin/rpm", "-q", "-f", path)
if outp, err := cmd.Output(); err == nil {
output = string(outp)
}
return strings.Trim(output, "\n")
}
func dpkgVersion(path string) string {
output := unknownPackage
cmd := exec.Command("/usr/bin/dpkg", "-S", path)
if outp, err := cmd.Output(); err == nil {
output = string(outp)
}
return strings.Trim(output, "\n")
}
func (r *OCIRuntime) pathPackage() string {
if out := rpmVersion(r.path); out != unknownPackage {
return out
}
return dpkgVersion(r.path)
}
func (r *OCIRuntime) conmonPackage() string {
if out := rpmVersion(r.conmonPath); out != unknownPackage {
return out
}
return dpkgVersion(r.conmonPath)
}
|
