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
|
// Copyright 2018 Google Inc.
//
// 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 host
import (
"os"
"path"
"syscall"
"gvisor.googlesource.com/gvisor/pkg/abi/linux"
"gvisor.googlesource.com/gvisor/pkg/log"
"gvisor.googlesource.com/gvisor/pkg/sentry/device"
"gvisor.googlesource.com/gvisor/pkg/sentry/fs"
"gvisor.googlesource.com/gvisor/pkg/sentry/kernel/auth"
ktime "gvisor.googlesource.com/gvisor/pkg/sentry/kernel/time"
"gvisor.googlesource.com/gvisor/pkg/syserror"
)
func open(parent *inodeOperations, name string) (int, error) {
if parent == nil && !path.IsAbs(name) {
return -1, syserror.EINVAL
}
name = path.Clean(name)
// Don't follow through symlinks.
flags := syscall.O_NOFOLLOW
if fd, err := openAt(parent, name, flags|syscall.O_RDWR, 0); err == nil {
return fd, nil
}
// Retry as read-only.
if fd, err := openAt(parent, name, flags|syscall.O_RDONLY, 0); err == nil {
return fd, nil
}
// Retry as write-only.
if fd, err := openAt(parent, name, flags|syscall.O_WRONLY, 0); err == nil {
return fd, nil
}
// Retry as a symlink, by including O_PATH as an option.
fd, err := openAt(parent, name, linux.O_PATH|flags, 0)
if err == nil {
return fd, nil
}
// Everything failed.
return -1, err
}
func openAt(parent *inodeOperations, name string, flags int, perm linux.FileMode) (int, error) {
if parent == nil {
return syscall.Open(name, flags, uint32(perm))
}
return syscall.Openat(parent.fileState.FD(), name, flags, uint32(perm))
}
func nodeType(s *syscall.Stat_t) fs.InodeType {
switch x := (s.Mode & syscall.S_IFMT); x {
case syscall.S_IFLNK:
return fs.Symlink
case syscall.S_IFIFO:
return fs.Pipe
case syscall.S_IFCHR:
return fs.CharacterDevice
case syscall.S_IFBLK:
return fs.BlockDevice
case syscall.S_IFSOCK:
return fs.Socket
case syscall.S_IFDIR:
return fs.Directory
case syscall.S_IFREG:
return fs.RegularFile
default:
// This shouldn't happen, but just in case...
log.Warningf("unknown host file type %d: assuming regular", x)
return fs.RegularFile
}
}
func wouldBlock(s *syscall.Stat_t) bool {
typ := nodeType(s)
return typ == fs.Pipe || typ == fs.Socket || typ == fs.CharacterDevice
}
func stableAttr(s *syscall.Stat_t) fs.StableAttr {
return fs.StableAttr{
Type: nodeType(s),
DeviceID: hostFileDevice.DeviceID(),
InodeID: hostFileDevice.Map(device.MultiDeviceKey{
Device: s.Dev,
Inode: s.Ino,
}),
BlockSize: int64(s.Blksize),
}
}
func owner(mo *superOperations, s *syscall.Stat_t) fs.FileOwner {
// User requested no translation, just return actual owner.
if mo.dontTranslateOwnership {
return fs.FileOwner{auth.KUID(s.Uid), auth.KGID(s.Gid)}
}
// Show only IDs relevant to the sandboxed task. I.e. if we not own the
// file, no sandboxed task can own the file. In that case, we
// use OverflowID for UID, implying that the IDs are not mapped in the
// "root" user namespace.
//
// E.g.
// sandbox's host EUID/EGID is 1/1.
// some_dir's host UID/GID is 2/1.
// Task that mounted this fs has virtualized EUID/EGID 5/5.
//
// If you executed `ls -n` in the sandboxed task, it would show:
// drwxwrxwrx [...] 65534 5 [...] some_dir
// Files are owned by OverflowID by default.
owner := fs.FileOwner{auth.KUID(auth.OverflowUID), auth.KGID(auth.OverflowGID)}
// If we own file on host, let mounting task's initial EUID own
// the file.
if s.Uid == hostUID {
owner.UID = mo.mounter.UID
}
// If our group matches file's group, make file's group match
// the mounting task's initial EGID.
for _, gid := range hostGIDs {
if s.Gid == gid {
owner.GID = mo.mounter.GID
break
}
}
return owner
}
func unstableAttr(mo *superOperations, s *syscall.Stat_t) fs.UnstableAttr {
return fs.UnstableAttr{
Size: s.Size,
Usage: s.Blocks * 512,
Perms: fs.FilePermsFromMode(linux.FileMode(s.Mode)),
Owner: owner(mo, s),
AccessTime: ktime.FromUnix(s.Atim.Sec, s.Atim.Nsec),
ModificationTime: ktime.FromUnix(s.Mtim.Sec, s.Mtim.Nsec),
StatusChangeTime: ktime.FromUnix(s.Ctim.Sec, s.Ctim.Nsec),
Links: s.Nlink,
}
}
type dirInfo struct {
buf []byte // buffer for directory I/O.
nbuf int // length of buf; return value from ReadDirent.
bufp int // location of next record in buf.
}
// isBlockError unwraps os errors and checks if they are caused by EAGAIN or
// EWOULDBLOCK. This is so they can be transformed into syserror.ErrWouldBlock.
func isBlockError(err error) bool {
if err == syserror.EAGAIN || err == syserror.EWOULDBLOCK {
return true
}
if pe, ok := err.(*os.PathError); ok {
return isBlockError(pe.Err)
}
return false
}
func hostEffectiveKIDs() (uint32, []uint32, error) {
gids, err := os.Getgroups()
if err != nil {
return 0, nil, err
}
egids := make([]uint32, len(gids))
for i, gid := range gids {
egids[i] = uint32(gid)
}
return uint32(os.Geteuid()), append(egids, uint32(os.Getegid())), nil
}
var hostUID uint32
var hostGIDs []uint32
func init() {
hostUID, hostGIDs, _ = hostEffectiveKIDs()
}
|
