summaryrefslogtreecommitdiff
path: root/Documentation/admin-guide/perf-security.rst
blob: 1307b5274a0f93e5650b9c31f5ee6908aebfefc3 (plain)
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
.. _perf_security:

Perf events and tool security
=============================

Overview
--------

Usage of Performance Counters for Linux (perf_events) [1]_ , [2]_ , [3]_
can impose a considerable risk of leaking sensitive data accessed by
monitored processes. The data leakage is possible both in scenarios of
direct usage of perf_events system call API [2]_ and over data files
generated by Perf tool user mode utility (Perf) [3]_ , [4]_ . The risk
depends on the nature of data that perf_events performance monitoring
units (PMU) [2]_ and Perf collect and expose for performance analysis.
Collected system and performance data may be split into several
categories:

1. System hardware and software configuration data, for example: a CPU
   model and its cache configuration, an amount of available memory and
   its topology, used kernel and Perf versions, performance monitoring
   setup including experiment time, events configuration, Perf command
   line parameters, etc.

2. User and kernel module paths and their load addresses with sizes,
   process and thread names with their PIDs and TIDs, timestamps for
   captured hardware and software events.

3. Content of kernel software counters (e.g., for context switches, page
   faults, CPU migrations), architectural hardware performance counters
   (PMC) [8]_ and machine specific registers (MSR) [9]_ that provide
   execution metrics for various monitored parts of the system (e.g.,
   memory controller (IMC), interconnect (QPI/UPI) or peripheral (PCIe)
   uncore counters) without direct attribution to any execution context
   state.

4. Content of architectural execution context registers (e.g., RIP, RSP,
   RBP on x86_64), process user and kernel space memory addresses and
   data, content of various architectural MSRs that capture data from
   this category.

Data that belong to the fourth category can potentially contain
sensitive process data. If PMUs in some monitoring modes capture values
of execution context registers or data from process memory then access
to such monitoring modes requires to be ordered and secured properly.
So, perf_events performance monitoring and observability operations are
the subject for security access control management [5]_ .

perf_events access control
-------------------------------

To perform security checks, the Linux implementation splits processes
into two categories [6]_ : a) privileged processes (whose effective user
ID is 0, referred to as superuser or root), and b) unprivileged
processes (whose effective UID is nonzero). Privileged processes bypass
all kernel security permission checks so perf_events performance
monitoring is fully available to privileged processes without access,
scope and resource restrictions.

Unprivileged processes are subject to a full security permission check
based on the process's credentials [5]_ (usually: effective UID,
effective GID, and supplementary group list).

Linux divides the privileges traditionally associated with superuser
into distinct units, known as capabilities [6]_ , which can be
independently enabled and disabled on per-thread basis for processes and
files of unprivileged users.

Unprivileged processes with enabled CAP_PERFMON capability are treated
as privileged processes with respect to perf_events performance
monitoring and observability operations, thus, bypass *scope* permissions
checks in the kernel. CAP_PERFMON implements the principle of least
privilege [13]_ (POSIX 1003.1e: 2.2.2.39) for performance monitoring and
observability operations in the kernel and provides a secure approach to
perfomance monitoring and observability in the system.

For backward compatibility reasons the access to perf_events monitoring and
observability operations is also open for CAP_SYS_ADMIN privileged
processes but CAP_SYS_ADMIN usage for secure monitoring and observability
use cases is discouraged with respect to the CAP_PERFMON capability.
If system audit records [14]_ for a process using perf_events system call
API contain denial records of acquiring both CAP_PERFMON and CAP_SYS_ADMIN
capabilities then providing the process with CAP_PERFMON capability singly
is recommended as the preferred secure approach to resolve double access
denial logging related to usage of performance monitoring and observability.

Unprivileged processes using perf_events system call are also subject
for PTRACE_MODE_READ_REALCREDS ptrace access mode check [7]_ , whose
outcome determines whether monitoring is permitted. So unprivileged
processes provided with CAP_SYS_PTRACE capability are effectively
permitted to pass the check.

Other capabilities being granted to unprivileged processes can
effectively enable capturing of additional data required for later
performance analysis of monitored processes or a system. For example,
CAP_SYSLOG capability permits reading kernel space memory addresses from
/proc/kallsyms file.

Privileged Perf users groups
---------------------------------

Mechanisms of capabilities, privileged capability-dumb files [6]_ and
file system ACLs [10]_ can be used to create dedicated groups of
privileged Perf users who are permitted to execute performance monitoring
and observability without scope limits. The following steps can be
taken to create such groups of privileged Perf users.

1. Create perf_users group of privileged Perf users, assign perf_users
   group to Perf tool executable and limit access to the executable for
   other users in the system who are not in the perf_users group:

::

   # groupadd perf_users
   # ls -alhF
   -rwxr-xr-x  2 root root  11M Oct 19 15:12 perf
   # chgrp perf_users perf
   # ls -alhF
   -rwxr-xr-x  2 root perf_users  11M Oct 19 15:12 perf
   # chmod o-rwx perf
   # ls -alhF
   -rwxr-x---  2 root perf_users  11M Oct 19 15:12 perf

2. Assign the required capabilities to the Perf tool executable file and
   enable members of perf_users group with monitoring and observability
   privileges [6]_ :

::

   # setcap "cap_perfmon,cap_sys_ptrace,cap_syslog=ep" perf
   # setcap -v "cap_perfmon,cap_sys_ptrace,cap_syslog=ep" perf
   perf: OK
   # getcap perf
   perf = cap_sys_ptrace,cap_syslog,cap_perfmon+ep

If the libcap installed doesn't yet support "cap_perfmon", use "38" instead,
i.e.:

::

   # setcap "38,cap_ipc_lock,cap_sys_ptrace,cap_syslog=ep" perf

Note that you may need to have 'cap_ipc_lock' in the mix for tools such as
'perf top', alternatively use 'perf top -m N', to reduce the memory that
it uses for the perf ring buffer, see the memory allocation section below.

Using a libcap without support for CAP_PERFMON will make cap_get_flag(caps, 38,
CAP_EFFECTIVE, &val) fail, which will lead the default event to be 'cycles:u',
so as a workaround explicitly ask for the 'cycles' event, i.e.:

::

  # perf top -e cycles

To get kernel and user samples with a perf binary with just CAP_PERFMON.

As a result, members of perf_users group are capable of conducting
performance monitoring and observability by using functionality of the
configured Perf tool executable that, when executes, passes perf_events
subsystem scope checks.

This specific access control management is only available to superuser
or root running processes with CAP_SETPCAP, CAP_SETFCAP [6]_
capabilities.

Unprivileged users
-----------------------------------

perf_events *scope* and *access* control for unprivileged processes
is governed by perf_event_paranoid [2]_ setting:

-1:
     Impose no *scope* and *access* restrictions on using perf_events
     performance monitoring. Per-user per-cpu perf_event_mlock_kb [2]_
     locking limit is ignored when allocating memory buffers for storing
     performance data. This is the least secure mode since allowed
     monitored *scope* is maximized and no perf_events specific limits
     are imposed on *resources* allocated for performance monitoring.

>=0:
     *scope* includes per-process and system wide performance monitoring
     but excludes raw tracepoints and ftrace function tracepoints
     monitoring. CPU and system events happened when executing either in
     user or in kernel space can be monitored and captured for later
     analysis. Per-user per-cpu perf_event_mlock_kb locking limit is
     imposed but ignored for unprivileged processes with CAP_IPC_LOCK
     [6]_ capability.

>=1:
     *scope* includes per-process performance monitoring only and
     excludes system wide performance monitoring. CPU and system events
     happened when executing either in user or in kernel space can be
     monitored and captured for later analysis. Per-user per-cpu
     perf_event_mlock_kb locking limit is imposed but ignored for
     unprivileged processes with CAP_IPC_LOCK capability.

>=2:
     *scope* includes per-process performance monitoring only. CPU and
     system events happened when executing in user space only can be
     monitored and captured for later analysis. Per-user per-cpu
     perf_event_mlock_kb locking limit is imposed but ignored for
     unprivileged processes with CAP_IPC_LOCK capability.

Resource control
---------------------------------

Open file descriptors
+++++++++++++++++++++

The perf_events system call API [2]_ allocates file descriptors for
every configured PMU event. Open file descriptors are a per-process
accountable resource governed by the RLIMIT_NOFILE [11]_ limit
(ulimit -n), which is usually derived from the login shell process. When
configuring Perf collection for a long list of events on a large server
system, this limit can be easily hit preventing required monitoring
configuration. RLIMIT_NOFILE limit can be increased on per-user basis
modifying content of the limits.conf file [12]_ . Ordinarily, a Perf
sampling session (perf record) requires an amount of open perf_event
file descriptors that is not less than the number of monitored events
multiplied by the number of monitored CPUs.

Memory allocation
+++++++++++++++++

The amount of memory available to user processes for capturing
performance monitoring data is governed by the perf_event_mlock_kb [2]_
setting. This perf_event specific resource setting defines overall
per-cpu limits of memory allowed for mapping by the user processes to
execute performance monitoring. The setting essentially extends the
RLIMIT_MEMLOCK [11]_ limit, but only for memory regions mapped
specifically for capturing monitored performance events and related data.

For example, if a machine has eight cores and perf_event_mlock_kb limit
is set to 516 KiB, then a user process is provided with 516 KiB * 8 =
4128 KiB of memory above the RLIMIT_MEMLOCK limit (ulimit -l) for
perf_event mmap buffers. In particular, this means that, if the user
wants to start two or more performance monitoring processes, the user is
required to manually distribute the available 4128 KiB between the
monitoring processes, for example, using the --mmap-pages Perf record
mode option. Otherwise, the first started performance monitoring process
allocates all available 4128 KiB and the other processes will fail to
proceed due to the lack of memory.

RLIMIT_MEMLOCK and perf_event_mlock_kb resource constraints are ignored
for processes with the CAP_IPC_LOCK capability. Thus, perf_events/Perf
privileged users can be provided with memory above the constraints for
perf_events/Perf performance monitoring purpose by providing the Perf
executable with CAP_IPC_LOCK capability.

Bibliography
------------

.. [1] `<https://lwn.net/Articles/337493/>`_
.. [2] `<http://man7.org/linux/man-pages/man2/perf_event_open.2.html>`_
.. [3] `<http://web.eece.maine.edu/~vweaver/projects/perf_events/>`_
.. [4] `<https://perf.wiki.kernel.org/index.php/Main_Page>`_
.. [5] `<https://www.kernel.org/doc/html/latest/security/credentials.html>`_
.. [6] `<http://man7.org/linux/man-pages/man7/capabilities.7.html>`_
.. [7] `<http://man7.org/linux/man-pages/man2/ptrace.2.html>`_
.. [8] `<https://en.wikipedia.org/wiki/Hardware_performance_counter>`_
.. [9] `<https://en.wikipedia.org/wiki/Model-specific_register>`_
.. [10] `<http://man7.org/linux/man-pages/man5/acl.5.html>`_
.. [11] `<http://man7.org/linux/man-pages/man2/getrlimit.2.html>`_
.. [12] `<http://man7.org/linux/man-pages/man5/limits.conf.5.html>`_
.. [13] `<https://sites.google.com/site/fullycapable>`_
.. [14] `<http://man7.org/linux/man-pages/man8/auditd.8.html>`_