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
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
|
// Copyright 2020 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"
"fmt"
"sort"
"strconv"
"gvisor.dev/gvisor/pkg/abi/linux"
"gvisor.dev/gvisor/pkg/context"
"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"
"gvisor.dev/gvisor/pkg/syserror"
)
func getTaskFD(t *kernel.Task, fd int32) (*vfs.FileDescription, kernel.FDFlags) {
var (
file *vfs.FileDescription
flags kernel.FDFlags
)
t.WithMuLocked(func(t *kernel.Task) {
if fdt := t.FDTable(); fdt != nil {
file, flags = fdt.GetVFS2(fd)
}
})
return file, flags
}
func taskFDExists(ctx context.Context, fs *filesystem, t *kernel.Task, fd int32) bool {
file, _ := getTaskFD(t, fd)
if file == nil {
return false
}
fs.SafeDecRefFD(ctx, file)
return true
}
// +stateify savable
type fdDir struct {
locks vfs.FileLocks
fs *filesystem
task *kernel.Task
// When produceSymlinks is set, dirents produces for the FDs are reported
// as symlink. Otherwise, they are reported as regular files.
produceSymlink bool
}
// IterDirents implements kernfs.inodeDirectory.IterDirents.
func (i *fdDir) IterDirents(ctx context.Context, mnt *vfs.Mount, cb vfs.IterDirentsCallback, offset, relOffset int64) (int64, error) {
var fds []int32
i.task.WithMuLocked(func(t *kernel.Task) {
if fdTable := t.FDTable(); fdTable != nil {
fds = fdTable.GetFDs(ctx)
}
})
typ := uint8(linux.DT_REG)
if i.produceSymlink {
typ = linux.DT_LNK
}
// Find the appropriate starting point.
idx := sort.Search(len(fds), func(i int) bool { return fds[i] >= int32(relOffset) })
if idx >= len(fds) {
return offset, nil
}
for _, fd := range fds[idx:] {
dirent := vfs.Dirent{
Name: strconv.FormatUint(uint64(fd), 10),
Type: typ,
Ino: i.fs.NextIno(),
NextOff: int64(fd) + 3,
}
if err := cb.Handle(dirent); err != nil {
// Getdents should iterate correctly despite mutation
// of fds, so we return the next fd to serialize plus
// 2 (which accounts for the "." and ".." tracked by
// kernfs) as the offset.
return int64(fd) + 2, err
}
}
// We serialized them all. Next offset should be higher than last
// serialized fd.
return int64(fds[len(fds)-1]) + 3, nil
}
// fdDirInode represents the inode for /proc/[pid]/fd directory.
//
// +stateify savable
type fdDirInode struct {
fdDir
fdDirInodeRefs
implStatFS
kernfs.InodeAlwaysValid
kernfs.InodeAttrs
kernfs.InodeDirectoryNoNewChildren
kernfs.InodeNotSymlink
kernfs.InodeTemporary
kernfs.OrderedChildren
}
var _ kernfs.Inode = (*fdDirInode)(nil)
func (fs *filesystem) newFDDirInode(ctx context.Context, task *kernel.Task) kernfs.Inode {
inode := &fdDirInode{
fdDir: fdDir{
fs: fs,
task: task,
produceSymlink: true,
},
}
inode.InodeAttrs.Init(ctx, task.Credentials(), linux.UNNAMED_MAJOR, fs.devMinor, fs.NextIno(), linux.ModeDirectory|0555)
inode.InitRefs()
inode.OrderedChildren.Init(kernfs.OrderedChildrenOptions{})
return inode
}
// IterDirents implements kernfs.inodeDirectory.IterDirents.
func (i *fdDirInode) IterDirents(ctx context.Context, mnt *vfs.Mount, cb vfs.IterDirentsCallback, offset, relOffset int64) (int64, error) {
return i.fdDir.IterDirents(ctx, mnt, cb, offset, relOffset)
}
// Lookup implements kernfs.inodeDirectory.Lookup.
func (i *fdDirInode) Lookup(ctx context.Context, name string) (kernfs.Inode, error) {
fdInt, err := strconv.ParseInt(name, 10, 32)
if err != nil {
return nil, syserror.ENOENT
}
fd := int32(fdInt)
if !taskFDExists(ctx, i.fs, i.task, fd) {
return nil, syserror.ENOENT
}
return i.fs.newFDSymlink(ctx, i.task, fd, i.fs.NextIno()), nil
}
// Open implements kernfs.Inode.Open.
func (i *fdDirInode) 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
}
// CheckPermissions implements kernfs.Inode.CheckPermissions.
//
// This is to match Linux, which uses a special permission handler to guarantee
// that a process can still access /proc/self/fd after it has executed
// setuid. See fs/proc/fd.c:proc_fd_permission.
func (i *fdDirInode) CheckPermissions(ctx context.Context, creds *auth.Credentials, ats vfs.AccessTypes) error {
err := i.InodeAttrs.CheckPermissions(ctx, creds, ats)
if err == nil {
// Access granted, no extra check needed.
return nil
}
if t := kernel.TaskFromContext(ctx); t != nil {
// Allow access if the task trying to access it is in the thread group
// corresponding to this directory.
if i.task.ThreadGroup() == t.ThreadGroup() {
// Access granted (overridden).
return nil
}
}
return err
}
// DecRef implements kernfs.Inode.DecRef.
func (i *fdDirInode) DecRef(ctx context.Context) {
i.fdDirInodeRefs.DecRef(func() { i.Destroy(ctx) })
}
// fdSymlink is an symlink for the /proc/[pid]/fd/[fd] file.
//
// +stateify savable
type fdSymlink struct {
implStatFS
kernfs.InodeAttrs
kernfs.InodeNoopRefCount
kernfs.InodeSymlink
fs *filesystem
task *kernel.Task
fd int32
}
var _ kernfs.Inode = (*fdSymlink)(nil)
func (fs *filesystem) newFDSymlink(ctx context.Context, task *kernel.Task, fd int32, ino uint64) kernfs.Inode {
inode := &fdSymlink{
fs: fs,
task: task,
fd: fd,
}
inode.Init(ctx, task.Credentials(), linux.UNNAMED_MAJOR, fs.devMinor, ino, linux.ModeSymlink|0777)
return inode
}
func (s *fdSymlink) Readlink(ctx context.Context, _ *vfs.Mount) (string, error) {
file, _ := getTaskFD(s.task, s.fd)
if file == nil {
return "", syserror.ENOENT
}
defer s.fs.SafeDecRefFD(ctx, file)
root := vfs.RootFromContext(ctx)
defer s.fs.SafeDecRef(ctx, root)
// Note: it's safe to reenter kernfs from Readlink if needed to resolve path.
return s.task.Kernel().VFS().PathnameWithDeleted(ctx, root, file.VirtualDentry())
}
func (s *fdSymlink) Getlink(ctx context.Context, mnt *vfs.Mount) (vfs.VirtualDentry, string, error) {
file, _ := getTaskFD(s.task, s.fd)
if file == nil {
return vfs.VirtualDentry{}, "", syserror.ENOENT
}
defer s.fs.SafeDecRefFD(ctx, file)
vd := file.VirtualDentry()
vd.IncRef()
return vd, "", nil
}
// Valid implements kernfs.Inode.Valid.
func (s *fdSymlink) Valid(ctx context.Context) bool {
return taskFDExists(ctx, s.fs, s.task, s.fd)
}
// fdInfoDirInode represents the inode for /proc/[pid]/fdinfo directory.
//
// +stateify savable
type fdInfoDirInode struct {
fdDir
fdInfoDirInodeRefs
implStatFS
kernfs.InodeAlwaysValid
kernfs.InodeAttrs
kernfs.InodeDirectoryNoNewChildren
kernfs.InodeNotSymlink
kernfs.InodeTemporary
kernfs.OrderedChildren
}
var _ kernfs.Inode = (*fdInfoDirInode)(nil)
func (fs *filesystem) newFDInfoDirInode(ctx context.Context, task *kernel.Task) kernfs.Inode {
inode := &fdInfoDirInode{
fdDir: fdDir{
fs: fs,
task: task,
},
}
inode.InodeAttrs.Init(ctx, task.Credentials(), linux.UNNAMED_MAJOR, fs.devMinor, fs.NextIno(), linux.ModeDirectory|0555)
inode.InitRefs()
inode.OrderedChildren.Init(kernfs.OrderedChildrenOptions{})
return inode
}
// Lookup implements kernfs.inodeDirectory.Lookup.
func (i *fdInfoDirInode) Lookup(ctx context.Context, name string) (kernfs.Inode, error) {
fdInt, err := strconv.ParseInt(name, 10, 32)
if err != nil {
return nil, syserror.ENOENT
}
fd := int32(fdInt)
if !taskFDExists(ctx, i.fs, i.task, fd) {
return nil, syserror.ENOENT
}
data := &fdInfoData{
fs: i.fs,
task: i.task,
fd: fd,
}
return i.fs.newTaskOwnedInode(ctx, i.task, i.fs.NextIno(), 0444, data), nil
}
// IterDirents implements Inode.IterDirents.
func (i *fdInfoDirInode) IterDirents(ctx context.Context, mnt *vfs.Mount, cb vfs.IterDirentsCallback, offset, relOffset int64) (newOffset int64, err error) {
return i.fdDir.IterDirents(ctx, mnt, cb, offset, relOffset)
}
// Open implements kernfs.Inode.Open.
func (i *fdInfoDirInode) 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
}
// DecRef implements kernfs.Inode.DecRef.
func (i *fdInfoDirInode) DecRef(ctx context.Context) {
i.fdInfoDirInodeRefs.DecRef(func() { i.Destroy(ctx) })
}
// fdInfoData implements vfs.DynamicBytesSource for /proc/[pid]/fdinfo/[fd].
//
// +stateify savable
type fdInfoData struct {
kernfs.DynamicBytesFile
fs *filesystem
task *kernel.Task
fd int32
}
var _ dynamicInode = (*fdInfoData)(nil)
// Generate implements vfs.DynamicBytesSource.Generate.
func (d *fdInfoData) Generate(ctx context.Context, buf *bytes.Buffer) error {
file, descriptorFlags := getTaskFD(d.task, d.fd)
if file == nil {
return syserror.ENOENT
}
defer d.fs.SafeDecRefFD(ctx, file)
// TODO(b/121266871): Include pos, locks, and other data. For now we only
// have flags.
// See https://www.kernel.org/doc/Documentation/filesystems/proc.txt
flags := uint(file.StatusFlags()) | descriptorFlags.ToLinuxFileFlags()
fmt.Fprintf(buf, "flags:\t0%o\n", flags)
return nil
}
// Valid implements kernfs.Inode.Valid.
func (d *fdInfoData) Valid(ctx context.Context) bool {
return taskFDExists(ctx, d.fs, d.task, d.fd)
}
|