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
|
// 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 vfs2
import (
"gvisor.dev/gvisor/pkg/abi/linux"
"gvisor.dev/gvisor/pkg/sentry/arch"
"gvisor.dev/gvisor/pkg/sentry/kernel"
"gvisor.dev/gvisor/pkg/sentry/vfs"
"gvisor.dev/gvisor/pkg/syserror"
"gvisor.dev/gvisor/pkg/usermem"
)
// Mount implements Linux syscall mount(2).
func Mount(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
sourceAddr := args[0].Pointer()
targetAddr := args[1].Pointer()
typeAddr := args[2].Pointer()
flags := args[3].Uint64()
dataAddr := args[4].Pointer()
// For null-terminated strings related to mount(2), Linux copies in at most
// a page worth of data. See fs/namespace.c:copy_mount_string().
fsType, err := t.CopyInString(typeAddr, usermem.PageSize)
if err != nil {
return 0, nil, err
}
source, err := t.CopyInString(sourceAddr, usermem.PageSize)
if err != nil {
return 0, nil, err
}
targetPath, err := copyInPath(t, targetAddr)
if err != nil {
return 0, nil, err
}
data := ""
if dataAddr != 0 {
// In Linux, a full page is always copied in regardless of null
// character placement, and the address is passed to each file system.
// Most file systems always treat this data as a string, though, and so
// do all of the ones we implement.
data, err = t.CopyInString(dataAddr, usermem.PageSize)
if err != nil {
return 0, nil, err
}
}
// Ignore magic value that was required before Linux 2.4.
if flags&linux.MS_MGC_MSK == linux.MS_MGC_VAL {
flags = flags &^ linux.MS_MGC_MSK
}
// Must have CAP_SYS_ADMIN in the current mount namespace's associated user
// namespace.
creds := t.Credentials()
if !creds.HasCapabilityIn(linux.CAP_SYS_ADMIN, t.MountNamespaceVFS2().Owner) {
return 0, nil, syserror.EPERM
}
const unsupportedOps = linux.MS_REMOUNT | linux.MS_BIND |
linux.MS_SHARED | linux.MS_PRIVATE | linux.MS_SLAVE |
linux.MS_UNBINDABLE | linux.MS_MOVE
// Silently allow MS_NOSUID, since we don't implement set-id bits
// anyway.
const unsupportedFlags = linux.MS_NODIRATIME | linux.MS_STRICTATIME
// Linux just allows passing any flags to mount(2) - it won't fail when
// unknown or unsupported flags are passed. Since we don't implement
// everything, we fail explicitly on flags that are unimplemented.
if flags&(unsupportedOps|unsupportedFlags) != 0 {
return 0, nil, syserror.EINVAL
}
var opts vfs.MountOptions
if flags&linux.MS_NOATIME == linux.MS_NOATIME {
opts.Flags.NoATime = true
}
if flags&linux.MS_NOEXEC == linux.MS_NOEXEC {
opts.Flags.NoExec = true
}
if flags&linux.MS_NODEV == linux.MS_NODEV {
opts.Flags.NoDev = true
}
if flags&linux.MS_NOSUID == linux.MS_NOSUID {
opts.Flags.NoSUID = true
}
if flags&linux.MS_RDONLY == linux.MS_RDONLY {
opts.ReadOnly = true
}
opts.GetFilesystemOptions.Data = data
target, err := getTaskPathOperation(t, linux.AT_FDCWD, targetPath, disallowEmptyPath, nofollowFinalSymlink)
if err != nil {
return 0, nil, err
}
defer target.Release(t)
_, err = t.Kernel().VFS().MountAt(t, creds, source, &target.pop, fsType, &opts)
return 0, nil, err
}
// Umount2 implements Linux syscall umount2(2).
func Umount2(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
flags := args[1].Int()
// Must have CAP_SYS_ADMIN in the mount namespace's associated user
// namespace.
//
// Currently, this is always the init task's user namespace.
creds := t.Credentials()
if !creds.HasCapabilityIn(linux.CAP_SYS_ADMIN, t.MountNamespaceVFS2().Owner) {
return 0, nil, syserror.EPERM
}
const unsupported = linux.MNT_FORCE | linux.MNT_EXPIRE
if flags&unsupported != 0 {
return 0, nil, syserror.EINVAL
}
path, err := copyInPath(t, addr)
if err != nil {
return 0, nil, err
}
tpop, err := getTaskPathOperation(t, linux.AT_FDCWD, path, disallowEmptyPath, nofollowFinalSymlink)
if err != nil {
return 0, nil, err
}
defer tpop.Release(t)
opts := vfs.UmountOptions{
Flags: uint32(flags),
}
return 0, nil, t.Kernel().VFS().UmountAt(t, creds, &tpop.pop, &opts)
}
|
