forked from google/nsjail
-
Notifications
You must be signed in to change notification settings - Fork 0
/
cgroup2.cc
281 lines (238 loc) · 8.77 KB
/
cgroup2.cc
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
/*
nsjail - cgroup2 namespacing
-----------------------------------------
Copyright 2014 Google Inc. All Rights Reserved.
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.
*/
#include "cgroup2.h"
#include <errno.h>
#include <fcntl.h>
#include <limits.h>
#include <linux/magic.h>
#include <stdarg.h>
#include <stdio.h>
#include <string.h>
#include <sys/stat.h>
#include <sys/vfs.h>
#include <unistd.h>
#include <fstream>
#include <iostream>
#include <sstream>
#include "logs.h"
#include "util.h"
namespace cgroup2 {
static bool addPidToProcList(const std::string &cgroup_path, pid_t pid) {
std::string pid_str = std::to_string(pid);
LOG_D("Adding pid='%s' to cgroup.procs", pid_str.c_str());
if (!util::writeBufToFile((cgroup_path + "/cgroup.procs").c_str(), pid_str.c_str(),
pid_str.length(), O_WRONLY)) {
LOG_W("Could not update cgroup.procs");
return false;
}
return true;
}
static std::string getCgroupPath(nsjconf_t *nsjconf, pid_t pid) {
return nsjconf->cgroupv2_mount + "/NSJAIL." + std::to_string(pid);
}
static std::string getJailCgroupPath(nsjconf_t *nsjconf) {
return nsjconf->cgroupv2_mount + "/NSJAIL_SELF." + std::to_string(getpid());
}
static bool createCgroup(const std::string &cgroup_path, pid_t pid) {
LOG_D("Create '%s' for pid=%d", cgroup_path.c_str(), (int)pid);
if (mkdir(cgroup_path.c_str(), 0700) == -1 && errno != EEXIST) {
PLOG_W("mkdir('%s', 0700) failed", cgroup_path.c_str());
return false;
}
return true;
}
static bool moveSelfIntoChildCgroup(nsjconf_t *nsjconf) {
/*
* Move ourselves into another group to avoid the 'No internal processes' rule
* https://unix.stackexchange.com/a/713343
*/
std::string jail_cgroup_path = getJailCgroupPath(nsjconf);
LOG_I("nsjail is moving itself to a new child cgroup: %s\n", jail_cgroup_path.c_str());
RETURN_ON_FAILURE(createCgroup(jail_cgroup_path, getpid()));
RETURN_ON_FAILURE(addPidToProcList(jail_cgroup_path, 0));
return true;
}
static bool enableCgroupSubtree(nsjconf_t *nsjconf, const std::string &controller, pid_t pid) {
std::string cgroup_path = nsjconf->cgroupv2_mount;
LOG_D("Enable cgroup.subtree_control +'%s' to '%s' for pid=%d", controller.c_str(),
cgroup_path.c_str(), pid);
std::string val = "+" + controller;
/*
* Try once without moving the nsjail process and if that fails then try moving the nsjail
* process into a child cgroup before trying a second time.
*/
if (util::writeBufToFile((cgroup_path + "/cgroup.subtree_control").c_str(), val.c_str(),
val.length(), O_WRONLY, false)) {
return true;
}
if (errno == EBUSY) {
RETURN_ON_FAILURE(moveSelfIntoChildCgroup(nsjconf));
if (util::writeBufToFile((cgroup_path + "/cgroup.subtree_control").c_str(),
val.c_str(), val.length(), O_WRONLY)) {
return true;
}
}
LOG_E(
"Could not apply '%s' to cgroup.subtree_control in '%s'. nsjail MUST be run from root "
"and the cgroup mount path must refer to the root/host cgroup to use cgroupv2. If you "
"use Docker, you may need to run the container with --cgroupns=host so that nsjail can"
" access the host/root cgroupv2 hierarchy. An alternative is mounting (or remounting) "
"the cgroupv2 filesystem but using the flag is just simpler.",
val.c_str(), cgroup_path.c_str());
return false;
}
static bool writeToCgroup(
const std::string &cgroup_path, const std::string &resource, const std::string &value) {
LOG_I("Setting '%s' to '%s'", resource.c_str(), value.c_str());
if (!util::writeBufToFile(
(cgroup_path + "/" + resource).c_str(), value.c_str(), value.length(), O_WRONLY)) {
LOG_W("Could not update %s", resource.c_str());
return false;
}
return true;
}
static void removeCgroup(const std::string &cgroup_path) {
LOG_D("Remove '%s'", cgroup_path.c_str());
if (rmdir(cgroup_path.c_str()) == -1) {
PLOG_W("rmdir('%s') failed", cgroup_path.c_str());
}
}
static bool needMemoryController(nsjconf_t *nsjconf) {
/*
* Check if we need 'memory'
* This matches the check in initNsFromParentMem()
*/
ssize_t swap_max = nsjconf->cgroup_mem_swap_max;
if (nsjconf->cgroup_mem_memsw_max > (size_t)0) {
swap_max = nsjconf->cgroup_mem_memsw_max - nsjconf->cgroup_mem_max;
}
if (nsjconf->cgroup_mem_max == (size_t)0 && swap_max < (ssize_t)0) {
return false;
}
return true;
}
static bool needPidsController(nsjconf_t *nsjconf) {
return nsjconf->cgroup_pids_max != 0;
}
static bool needCpuController(nsjconf_t *nsjconf) {
return nsjconf->cgroup_cpu_ms_per_sec != 0U;
}
/*
* We will use this buf to read from cgroup.subtree_control to see if
* the root cgroup has the necessary controllers listed
*/
#define SUBTREE_CONTROL_BUF_LEN 0x40
bool setup(nsjconf_t *nsjconf) {
/*
* Read from cgroup.subtree_control in the root to see if
* the controllers we need are there.
*/
auto p = nsjconf->cgroupv2_mount + "/cgroup.subtree_control";
char buf[SUBTREE_CONTROL_BUF_LEN];
int read = util::readFromFile(p.c_str(), buf, SUBTREE_CONTROL_BUF_LEN - 1);
if (read < 0) {
LOG_W("cgroupv2 setup: Could not read root subtree_control");
return false;
}
buf[read] = 0;
/* Are the controllers we need there? */
bool subtree_ok = (!needMemoryController(nsjconf) || strstr(buf, "memory")) &&
(!needPidsController(nsjconf) || strstr(buf, "pids")) &&
(!needCpuController(nsjconf) || strstr(buf, "cpu"));
if (!subtree_ok) {
/* Now we can write to the root cgroup.subtree_control */
if (needMemoryController(nsjconf)) {
RETURN_ON_FAILURE(enableCgroupSubtree(nsjconf, "memory", getpid()));
}
if (needPidsController(nsjconf)) {
RETURN_ON_FAILURE(enableCgroupSubtree(nsjconf, "pids", getpid()));
}
if (needCpuController(nsjconf)) {
RETURN_ON_FAILURE(enableCgroupSubtree(nsjconf, "cpu", getpid()));
}
}
return true;
}
bool detectCgroupv2(nsjconf_t *nsjconf) {
/*
* Check cgroupv2_mount, if it is a cgroup2 mount, use it.
*/
struct statfs buf;
if (statfs(nsjconf->cgroupv2_mount.c_str(), &buf)) {
LOG_D("statfs %s failed with %d", nsjconf->cgroupv2_mount.c_str(), errno);
nsjconf->use_cgroupv2 = false;
return false;
}
nsjconf->use_cgroupv2 = (buf.f_type == CGROUP2_SUPER_MAGIC);
return true;
}
static bool initNsFromParentMem(nsjconf_t *nsjconf, pid_t pid) {
ssize_t swap_max = nsjconf->cgroup_mem_swap_max;
if (nsjconf->cgroup_mem_memsw_max > (size_t)0) {
swap_max = nsjconf->cgroup_mem_memsw_max - nsjconf->cgroup_mem_max;
}
if (nsjconf->cgroup_mem_max == (size_t)0 && swap_max < (ssize_t)0) {
return true;
}
std::string cgroup_path = getCgroupPath(nsjconf, pid);
RETURN_ON_FAILURE(createCgroup(cgroup_path, pid));
RETURN_ON_FAILURE(addPidToProcList(cgroup_path, pid));
if (nsjconf->cgroup_mem_max > (size_t)0) {
RETURN_ON_FAILURE(writeToCgroup(
cgroup_path, "memory.max", std::to_string(nsjconf->cgroup_mem_max)));
}
if (swap_max >= (ssize_t)0) {
RETURN_ON_FAILURE(
writeToCgroup(cgroup_path, "memory.swap.max", std::to_string(swap_max)));
}
return true;
}
static bool initNsFromParentPids(nsjconf_t *nsjconf, pid_t pid) {
if (nsjconf->cgroup_pids_max == 0U) {
return true;
}
std::string cgroup_path = getCgroupPath(nsjconf, pid);
RETURN_ON_FAILURE(createCgroup(cgroup_path, pid));
RETURN_ON_FAILURE(addPidToProcList(cgroup_path, pid));
return writeToCgroup(cgroup_path, "pids.max", std::to_string(nsjconf->cgroup_pids_max));
}
static bool initNsFromParentCpu(nsjconf_t *nsjconf, pid_t pid) {
if (nsjconf->cgroup_cpu_ms_per_sec == 0U) {
return true;
}
std::string cgroup_path = getCgroupPath(nsjconf, pid);
RETURN_ON_FAILURE(createCgroup(cgroup_path, pid));
RETURN_ON_FAILURE(addPidToProcList(cgroup_path, pid));
/*
* The maximum bandwidth limit in the format: `$MAX $PERIOD`.
* This indicates that the group may consume up to $MAX in each $PERIOD
* duration.
*/
std::string cpu_ms_per_sec_str = std::to_string(nsjconf->cgroup_cpu_ms_per_sec * 1000U);
cpu_ms_per_sec_str += " 1000000";
return writeToCgroup(cgroup_path, "cpu.max", cpu_ms_per_sec_str);
}
bool initNsFromParent(nsjconf_t *nsjconf, pid_t pid) {
RETURN_ON_FAILURE(initNsFromParentMem(nsjconf, pid));
RETURN_ON_FAILURE(initNsFromParentPids(nsjconf, pid));
return initNsFromParentCpu(nsjconf, pid);
}
void finishFromParent(nsjconf_t *nsjconf, pid_t pid) {
if (nsjconf->cgroup_mem_max != (size_t)0 || nsjconf->cgroup_pids_max != 0U ||
nsjconf->cgroup_cpu_ms_per_sec != 0U) {
removeCgroup(getCgroupPath(nsjconf, pid));
}
}
} // namespace cgroup2