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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271
|
// Copyright 2019 The gVisor Authors.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package proc
import (
"bytes"
"sort"
"strconv"
"gvisor.dev/gvisor/pkg/abi/linux"
"gvisor.dev/gvisor/pkg/context"
"gvisor.dev/gvisor/pkg/errors/linuxerr"
"gvisor.dev/gvisor/pkg/sentry/fsimpl/kernfs"
"gvisor.dev/gvisor/pkg/sentry/kernel"
"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
"gvisor.dev/gvisor/pkg/sentry/vfs"
)
const (
selfName = "self"
threadSelfName = "thread-self"
)
// tasksInode represents the inode for /proc/ directory.
//
// +stateify savable
type tasksInode struct {
implStatFS
kernfs.InodeAlwaysValid
kernfs.InodeAttrs
kernfs.InodeDirectoryNoNewChildren
kernfs.InodeNotSymlink
kernfs.InodeTemporary // This holds no meaning as this inode can't be Looked up and is always valid.
kernfs.InodeWatches
kernfs.OrderedChildren
tasksInodeRefs
locks vfs.FileLocks
fs *filesystem
pidns *kernel.PIDNamespace
// '/proc/self' and '/proc/thread-self' have custom directory offsets in
// Linux. So handle them outside of OrderedChildren.
// fakeCgroupControllers is a map of controller name to directory in the
// cgroup hierarchy. These controllers are immutable and will be listed
// in /proc/pid/cgroup if not nil.
fakeCgroupControllers map[string]string
}
var _ kernfs.Inode = (*tasksInode)(nil)
func (fs *filesystem) newTasksInode(ctx context.Context, k *kernel.Kernel, pidns *kernel.PIDNamespace, fakeCgroupControllers map[string]string) *tasksInode {
root := auth.NewRootCredentials(pidns.UserNamespace())
contents := map[string]kernfs.Inode{
"cmdline": fs.newInode(ctx, root, 0444, &cmdLineData{}),
"cpuinfo": fs.newInode(ctx, root, 0444, newStaticFileSetStat(cpuInfoData(k))),
"filesystems": fs.newInode(ctx, root, 0444, &filesystemsData{}),
"loadavg": fs.newInode(ctx, root, 0444, &loadavgData{}),
"sys": fs.newSysDir(ctx, root, k),
"meminfo": fs.newInode(ctx, root, 0444, &meminfoData{}),
"mounts": kernfs.NewStaticSymlink(ctx, root, linux.UNNAMED_MAJOR, fs.devMinor, fs.NextIno(), "self/mounts"),
"net": kernfs.NewStaticSymlink(ctx, root, linux.UNNAMED_MAJOR, fs.devMinor, fs.NextIno(), "self/net"),
"sentry-meminfo": fs.newInode(ctx, root, 0444, &sentryMeminfoData{}),
"stat": fs.newInode(ctx, root, 0444, &statData{}),
"uptime": fs.newInode(ctx, root, 0444, &uptimeData{}),
"version": fs.newInode(ctx, root, 0444, &versionData{}),
}
// If fakeCgroupControllers are provided, don't create a cgroupfs backed
// /proc/cgroup as it will not match the fake controllers.
if len(fakeCgroupControllers) == 0 {
contents["cgroups"] = fs.newInode(ctx, root, 0444, &cgroupsData{})
}
inode := &tasksInode{
pidns: pidns,
fs: fs,
fakeCgroupControllers: fakeCgroupControllers,
}
inode.InodeAttrs.Init(ctx, root, linux.UNNAMED_MAJOR, fs.devMinor, fs.NextIno(), linux.ModeDirectory|0555)
inode.InitRefs()
inode.OrderedChildren.Init(kernfs.OrderedChildrenOptions{})
links := inode.OrderedChildren.Populate(contents)
inode.IncLinks(links)
return inode
}
// Lookup implements kernfs.inodeDirectory.Lookup.
func (i *tasksInode) Lookup(ctx context.Context, name string) (kernfs.Inode, error) {
// Check if a static entry was looked up.
if d, err := i.OrderedChildren.Lookup(ctx, name); err == nil {
return d, nil
}
// Not a static entry. Try to lookup a corresponding task.
tid, err := strconv.ParseUint(name, 10, 64)
if err != nil {
root := auth.NewRootCredentials(i.pidns.UserNamespace())
// If it failed to parse, check if it's one of the special handled files.
switch name {
case selfName:
return i.newSelfSymlink(ctx, root), nil
case threadSelfName:
return i.newThreadSelfSymlink(ctx, root), nil
}
return nil, linuxerr.ENOENT
}
task := i.pidns.TaskWithID(kernel.ThreadID(tid))
if task == nil {
return nil, linuxerr.ENOENT
}
return i.fs.newTaskInode(ctx, task, i.pidns, true, i.fakeCgroupControllers)
}
// IterDirents implements kernfs.inodeDirectory.IterDirents.
func (i *tasksInode) IterDirents(ctx context.Context, mnt *vfs.Mount, cb vfs.IterDirentsCallback, offset, _ int64) (int64, error) {
// fs/proc/internal.h: #define FIRST_PROCESS_ENTRY 256
const FIRST_PROCESS_ENTRY = 256
// Use maxTaskID to shortcut searches that will result in 0 entries.
const maxTaskID = kernel.TasksLimit + 1
if offset >= maxTaskID {
return offset, nil
}
// According to Linux (fs/proc/base.c:proc_pid_readdir()), process directories
// start at offset FIRST_PROCESS_ENTRY with '/proc/self', followed by
// '/proc/thread-self' and then '/proc/[pid]'.
if offset < FIRST_PROCESS_ENTRY {
offset = FIRST_PROCESS_ENTRY
}
if offset == FIRST_PROCESS_ENTRY {
dirent := vfs.Dirent{
Name: selfName,
Type: linux.DT_LNK,
Ino: i.fs.NextIno(),
NextOff: offset + 1,
}
if err := cb.Handle(dirent); err != nil {
return offset, err
}
offset++
}
if offset == FIRST_PROCESS_ENTRY+1 {
dirent := vfs.Dirent{
Name: threadSelfName,
Type: linux.DT_LNK,
Ino: i.fs.NextIno(),
NextOff: offset + 1,
}
if err := cb.Handle(dirent); err != nil {
return offset, err
}
offset++
}
// Collect all tasks that TGIDs are greater than the offset specified. Per
// Linux we only include in directory listings if it's the leader. But for
// whatever crazy reason, you can still walk to the given node.
var tids []int
startTid := offset - FIRST_PROCESS_ENTRY - 2
for _, tg := range i.pidns.ThreadGroups() {
tid := i.pidns.IDOfThreadGroup(tg)
if int64(tid) < startTid {
continue
}
if leader := tg.Leader(); leader != nil {
tids = append(tids, int(tid))
}
}
if len(tids) == 0 {
return offset, nil
}
sort.Ints(tids)
for _, tid := range tids {
dirent := vfs.Dirent{
Name: strconv.FormatUint(uint64(tid), 10),
Type: linux.DT_DIR,
Ino: i.fs.NextIno(),
NextOff: FIRST_PROCESS_ENTRY + 2 + int64(tid) + 1,
}
if err := cb.Handle(dirent); err != nil {
return offset, err
}
offset++
}
return maxTaskID, nil
}
// Open implements kernfs.Inode.Open.
func (i *tasksInode) Open(ctx context.Context, rp *vfs.ResolvingPath, d *kernfs.Dentry, opts vfs.OpenOptions) (*vfs.FileDescription, error) {
fd, err := kernfs.NewGenericDirectoryFD(rp.Mount(), d, &i.OrderedChildren, &i.locks, &opts, kernfs.GenericDirectoryFDOptions{
SeekEnd: kernfs.SeekEndZero,
})
if err != nil {
return nil, err
}
return fd.VFSFileDescription(), nil
}
func (i *tasksInode) Stat(ctx context.Context, vsfs *vfs.Filesystem, opts vfs.StatOptions) (linux.Statx, error) {
stat, err := i.InodeAttrs.Stat(ctx, vsfs, opts)
if err != nil {
return linux.Statx{}, err
}
if opts.Mask&linux.STATX_NLINK != 0 {
// Add dynamic children to link count.
for _, tg := range i.pidns.ThreadGroups() {
if leader := tg.Leader(); leader != nil {
stat.Nlink++
}
}
}
return stat, nil
}
// DecRef implements kernfs.Inode.DecRef.
func (i *tasksInode) DecRef(ctx context.Context) {
i.tasksInodeRefs.DecRef(func() { i.Destroy(ctx) })
}
// staticFileSetStat implements a special static file that allows inode
// attributes to be set. This is to support /proc files that are readonly, but
// allow attributes to be set.
//
// +stateify savable
type staticFileSetStat struct {
dynamicBytesFileSetAttr
vfs.StaticData
}
var _ dynamicInode = (*staticFileSetStat)(nil)
func newStaticFileSetStat(data string) *staticFileSetStat {
return &staticFileSetStat{StaticData: vfs.StaticData{Data: data}}
}
func cpuInfoData(k *kernel.Kernel) string {
features := k.FeatureSet()
var buf bytes.Buffer
for i, max := uint(0), k.ApplicationCores(); i < max; i++ {
features.WriteCPUInfoTo(i, &buf)
}
return buf.String()
}
func ipcData(v uint64) dynamicInode {
return newStaticFile(strconv.FormatUint(v, 10))
}
|