1 # SPDX-License-Identifier: GPL-2.0-only 1 # SPDX-License-Identifier: GPL-2.0-only
2 menu "Kernel hardening options" 2 menu "Kernel hardening options"
3 3
4 config GCC_PLUGIN_STRUCTLEAK 4 config GCC_PLUGIN_STRUCTLEAK
5 bool 5 bool
6 help 6 help
7 While the kernel is built with warni 7 While the kernel is built with warnings enabled for any missed
8 stack variable initializations, this 8 stack variable initializations, this warning is silenced for
9 anything passed by reference to anot 9 anything passed by reference to another function, under the
10 occasionally misguided assumption th 10 occasionally misguided assumption that the function will do
11 the initialization. As this regularl 11 the initialization. As this regularly leads to exploitable
12 flaws, this plugin is available to i 12 flaws, this plugin is available to identify and zero-initialize
13 such variables, depending on the cho 13 such variables, depending on the chosen level of coverage.
14 14
15 This plugin was originally ported fr 15 This plugin was originally ported from grsecurity/PaX. More
16 information at: 16 information at:
17 * https://grsecurity.net/ 17 * https://grsecurity.net/
18 * https://pax.grsecurity.net/ 18 * https://pax.grsecurity.net/
19 19
20 menu "Memory initialization" 20 menu "Memory initialization"
21 21
22 config CC_HAS_AUTO_VAR_INIT_PATTERN 22 config CC_HAS_AUTO_VAR_INIT_PATTERN
23 def_bool $(cc-option,-ftrivial-auto-va 23 def_bool $(cc-option,-ftrivial-auto-var-init=pattern)
24 24
25 config CC_HAS_AUTO_VAR_INIT_ZERO_BARE <<
26 def_bool $(cc-option,-ftrivial-auto-va <<
27 <<
28 config CC_HAS_AUTO_VAR_INIT_ZERO_ENABLER <<
29 # Clang 16 and later warn about using <<
30 # is required before then. <<
31 def_bool $(cc-option,-ftrivial-auto-va <<
32 depends on !CC_HAS_AUTO_VAR_INIT_ZERO_ <<
33 <<
34 config CC_HAS_AUTO_VAR_INIT_ZERO 25 config CC_HAS_AUTO_VAR_INIT_ZERO
35 def_bool CC_HAS_AUTO_VAR_INIT_ZERO_BAR !! 26 # GCC ignores the -enable flag, so we can test for the feature with
>> 27 # a single invocation using the flag, but drop it as appropriate in
>> 28 # the Makefile, depending on the presence of Clang.
>> 29 def_bool $(cc-option,-ftrivial-auto-var-init=zero -enable-trivial-auto-var-init-zero-knowing-it-will-be-removed-from-clang)
36 30
37 choice 31 choice
38 prompt "Initialize kernel stack variab 32 prompt "Initialize kernel stack variables at function entry"
39 default GCC_PLUGIN_STRUCTLEAK_BYREF_AL 33 default GCC_PLUGIN_STRUCTLEAK_BYREF_ALL if COMPILE_TEST && GCC_PLUGINS
40 default INIT_STACK_ALL_PATTERN if COMP 34 default INIT_STACK_ALL_PATTERN if COMPILE_TEST && CC_HAS_AUTO_VAR_INIT_PATTERN
41 default INIT_STACK_ALL_ZERO if CC_HAS_ 35 default INIT_STACK_ALL_ZERO if CC_HAS_AUTO_VAR_INIT_ZERO
42 default INIT_STACK_NONE 36 default INIT_STACK_NONE
43 help 37 help
44 This option enables initialization o 38 This option enables initialization of stack variables at
45 function entry time. This has the po 39 function entry time. This has the possibility to have the
46 greatest coverage (since all functio 40 greatest coverage (since all functions can have their
47 variables initialized), but the perf 41 variables initialized), but the performance impact depends
48 on the function calling complexity o 42 on the function calling complexity of a given workload's
49 syscalls. 43 syscalls.
50 44
51 This chooses the level of coverage o 45 This chooses the level of coverage over classes of potentially
52 uninitialized variables. The selecte 46 uninitialized variables. The selected class of variable will be
53 initialized before use in a function 47 initialized before use in a function.
54 48
55 config INIT_STACK_NONE 49 config INIT_STACK_NONE
56 bool "no automatic stack varia 50 bool "no automatic stack variable initialization (weakest)"
57 help 51 help
58 Disable automatic stack vari 52 Disable automatic stack variable initialization.
59 This leaves the kernel vulne 53 This leaves the kernel vulnerable to the standard
60 classes of uninitialized sta 54 classes of uninitialized stack variable exploits
61 and information exposures. 55 and information exposures.
62 56
63 config GCC_PLUGIN_STRUCTLEAK_USER 57 config GCC_PLUGIN_STRUCTLEAK_USER
64 bool "zero-init structs marked 58 bool "zero-init structs marked for userspace (weak)"
65 # Plugin can be removed once t 59 # Plugin can be removed once the kernel only supports GCC 12+
66 depends on GCC_PLUGINS && !CC_ 60 depends on GCC_PLUGINS && !CC_HAS_AUTO_VAR_INIT_ZERO
67 select GCC_PLUGIN_STRUCTLEAK 61 select GCC_PLUGIN_STRUCTLEAK
68 help 62 help
69 Zero-initialize any structur 63 Zero-initialize any structures on the stack containing
70 a __user attribute. This can 64 a __user attribute. This can prevent some classes of
71 uninitialized stack variable 65 uninitialized stack variable exploits and information
72 exposures, like CVE-2013-214 66 exposures, like CVE-2013-2141:
73 https://git.kernel.org/linus 67 https://git.kernel.org/linus/b9e146d8eb3b9eca
74 68
75 config GCC_PLUGIN_STRUCTLEAK_BYREF 69 config GCC_PLUGIN_STRUCTLEAK_BYREF
76 bool "zero-init structs passed 70 bool "zero-init structs passed by reference (strong)"
77 # Plugin can be removed once t 71 # Plugin can be removed once the kernel only supports GCC 12+
78 depends on GCC_PLUGINS && !CC_ 72 depends on GCC_PLUGINS && !CC_HAS_AUTO_VAR_INIT_ZERO
79 depends on !(KASAN && KASAN_ST 73 depends on !(KASAN && KASAN_STACK)
80 select GCC_PLUGIN_STRUCTLEAK 74 select GCC_PLUGIN_STRUCTLEAK
81 help 75 help
82 Zero-initialize any structur 76 Zero-initialize any structures on the stack that may
83 be passed by reference and h 77 be passed by reference and had not already been
84 explicitly initialized. This 78 explicitly initialized. This can prevent most classes
85 of uninitialized stack varia 79 of uninitialized stack variable exploits and information
86 exposures, like CVE-2017-100 80 exposures, like CVE-2017-1000410:
87 https://git.kernel.org/linus 81 https://git.kernel.org/linus/06e7e776ca4d3654
88 82
89 As a side-effect, this keeps 83 As a side-effect, this keeps a lot of variables on the
90 stack that can otherwise be 84 stack that can otherwise be optimized out, so combining
91 this with CONFIG_KASAN_STACK 85 this with CONFIG_KASAN_STACK can lead to a stack overflow
92 and is disallowed. 86 and is disallowed.
93 87
94 config GCC_PLUGIN_STRUCTLEAK_BYREF_ALL 88 config GCC_PLUGIN_STRUCTLEAK_BYREF_ALL
95 bool "zero-init everything pas 89 bool "zero-init everything passed by reference (very strong)"
96 # Plugin can be removed once t 90 # Plugin can be removed once the kernel only supports GCC 12+
97 depends on GCC_PLUGINS && !CC_ 91 depends on GCC_PLUGINS && !CC_HAS_AUTO_VAR_INIT_ZERO
98 depends on !(KASAN && KASAN_ST 92 depends on !(KASAN && KASAN_STACK)
99 select GCC_PLUGIN_STRUCTLEAK 93 select GCC_PLUGIN_STRUCTLEAK
100 help 94 help
101 Zero-initialize any stack va 95 Zero-initialize any stack variables that may be passed
102 by reference and had not alr 96 by reference and had not already been explicitly
103 initialized. This is intende 97 initialized. This is intended to eliminate all classes
104 of uninitialized stack varia 98 of uninitialized stack variable exploits and information
105 exposures. 99 exposures.
106 100
107 As a side-effect, this keeps 101 As a side-effect, this keeps a lot of variables on the
108 stack that can otherwise be 102 stack that can otherwise be optimized out, so combining
109 this with CONFIG_KASAN_STACK 103 this with CONFIG_KASAN_STACK can lead to a stack overflow
110 and is disallowed. 104 and is disallowed.
111 105
112 config INIT_STACK_ALL_PATTERN 106 config INIT_STACK_ALL_PATTERN
113 bool "pattern-init everything 107 bool "pattern-init everything (strongest)"
114 depends on CC_HAS_AUTO_VAR_INI 108 depends on CC_HAS_AUTO_VAR_INIT_PATTERN
115 depends on !KMSAN <<
116 help 109 help
117 Initializes everything on th 110 Initializes everything on the stack (including padding)
118 with a specific debug value. 111 with a specific debug value. This is intended to eliminate
119 all classes of uninitialized 112 all classes of uninitialized stack variable exploits and
120 information exposures, even 113 information exposures, even variables that were warned about
121 having been left uninitializ 114 having been left uninitialized.
122 115
123 Pattern initialization is kn 116 Pattern initialization is known to provoke many existing bugs
124 related to uninitialized loc 117 related to uninitialized locals, e.g. pointers receive
125 non-NULL values, buffer size 118 non-NULL values, buffer sizes and indices are very big. The
126 pattern is situation-specifi 119 pattern is situation-specific; Clang on 64-bit uses 0xAA
127 repeating for all types and 120 repeating for all types and padding except float and double
128 which use 0xFF repeating (-N 121 which use 0xFF repeating (-NaN). Clang on 32-bit uses 0xFF
129 repeating for all types and 122 repeating for all types and padding.
130 123
131 config INIT_STACK_ALL_ZERO 124 config INIT_STACK_ALL_ZERO
132 bool "zero-init everything (st 125 bool "zero-init everything (strongest and safest)"
133 depends on CC_HAS_AUTO_VAR_INI 126 depends on CC_HAS_AUTO_VAR_INIT_ZERO
134 depends on !KMSAN <<
135 help 127 help
136 Initializes everything on th 128 Initializes everything on the stack (including padding)
137 with a zero value. This is i 129 with a zero value. This is intended to eliminate all
138 classes of uninitialized sta 130 classes of uninitialized stack variable exploits and
139 information exposures, even 131 information exposures, even variables that were warned
140 about having been left unini 132 about having been left uninitialized.
141 133
142 Zero initialization provides 134 Zero initialization provides safe defaults for strings
143 (immediately NUL-terminated) 135 (immediately NUL-terminated), pointers (NULL), indices
144 (index 0), and sizes (0 leng 136 (index 0), and sizes (0 length), so it is therefore more
145 suitable as a production sec 137 suitable as a production security mitigation than pattern
146 initialization. 138 initialization.
147 139
148 endchoice 140 endchoice
149 141
150 config GCC_PLUGIN_STRUCTLEAK_VERBOSE 142 config GCC_PLUGIN_STRUCTLEAK_VERBOSE
151 bool "Report forcefully initialized va 143 bool "Report forcefully initialized variables"
152 depends on GCC_PLUGIN_STRUCTLEAK 144 depends on GCC_PLUGIN_STRUCTLEAK
153 depends on !COMPILE_TEST # too 145 depends on !COMPILE_TEST # too noisy
154 help 146 help
155 This option will cause a warning to 147 This option will cause a warning to be printed each time the
156 structleak plugin finds a variable i 148 structleak plugin finds a variable it thinks needs to be
157 initialized. Since not all existing 149 initialized. Since not all existing initializers are detected
158 by the plugin, this can produce fals 150 by the plugin, this can produce false positive warnings.
159 151
160 config GCC_PLUGIN_STACKLEAK 152 config GCC_PLUGIN_STACKLEAK
161 bool "Poison kernel stack before retur 153 bool "Poison kernel stack before returning from syscalls"
162 depends on GCC_PLUGINS 154 depends on GCC_PLUGINS
163 depends on HAVE_ARCH_STACKLEAK 155 depends on HAVE_ARCH_STACKLEAK
164 help 156 help
165 This option makes the kernel erase t 157 This option makes the kernel erase the kernel stack before
166 returning from system calls. This ha 158 returning from system calls. This has the effect of leaving
167 the stack initialized to the poison 159 the stack initialized to the poison value, which both reduces
168 the lifetime of any sensitive stack 160 the lifetime of any sensitive stack contents and reduces
169 potential for uninitialized stack va 161 potential for uninitialized stack variable exploits or information
170 exposures (it does not cover functio 162 exposures (it does not cover functions reaching the same stack
171 depth as prior functions during the 163 depth as prior functions during the same syscall). This blocks
172 most uninitialized stack variable at 164 most uninitialized stack variable attacks, with the performance
173 impact being driven by the depth of 165 impact being driven by the depth of the stack usage, rather than
174 the function calling complexity. 166 the function calling complexity.
175 167
176 The performance impact on a single C 168 The performance impact on a single CPU system kernel compilation
177 sees a 1% slowdown, other systems an 169 sees a 1% slowdown, other systems and workloads may vary and you
178 are advised to test this feature on 170 are advised to test this feature on your expected workload before
179 deploying it. 171 deploying it.
180 172
181 This plugin was ported from grsecuri 173 This plugin was ported from grsecurity/PaX. More information at:
182 * https://grsecurity.net/ 174 * https://grsecurity.net/
183 * https://pax.grsecurity.net/ 175 * https://pax.grsecurity.net/
184 176
185 config GCC_PLUGIN_STACKLEAK_VERBOSE 177 config GCC_PLUGIN_STACKLEAK_VERBOSE
186 bool "Report stack depth analysis inst 178 bool "Report stack depth analysis instrumentation" if EXPERT
187 depends on GCC_PLUGIN_STACKLEAK 179 depends on GCC_PLUGIN_STACKLEAK
188 depends on !COMPILE_TEST # too 180 depends on !COMPILE_TEST # too noisy
189 help 181 help
190 This option will cause a warning to 182 This option will cause a warning to be printed each time the
191 stackleak plugin finds a function it 183 stackleak plugin finds a function it thinks needs to be
192 instrumented. This is useful for com 184 instrumented. This is useful for comparing coverage between
193 builds. 185 builds.
194 186
195 config STACKLEAK_TRACK_MIN_SIZE 187 config STACKLEAK_TRACK_MIN_SIZE
196 int "Minimum stack frame size of funct 188 int "Minimum stack frame size of functions tracked by STACKLEAK"
197 default 100 189 default 100
198 range 0 4096 190 range 0 4096
199 depends on GCC_PLUGIN_STACKLEAK 191 depends on GCC_PLUGIN_STACKLEAK
200 help 192 help
201 The STACKLEAK gcc plugin instruments 193 The STACKLEAK gcc plugin instruments the kernel code for tracking
202 the lowest border of the kernel stac 194 the lowest border of the kernel stack (and for some other purposes).
203 It inserts the stackleak_track_stack 195 It inserts the stackleak_track_stack() call for the functions with
204 a stack frame size greater than or e 196 a stack frame size greater than or equal to this parameter.
205 If unsure, leave the default value 1 197 If unsure, leave the default value 100.
206 198
207 config STACKLEAK_METRICS 199 config STACKLEAK_METRICS
208 bool "Show STACKLEAK metrics in the /p 200 bool "Show STACKLEAK metrics in the /proc file system"
209 depends on GCC_PLUGIN_STACKLEAK 201 depends on GCC_PLUGIN_STACKLEAK
210 depends on PROC_FS 202 depends on PROC_FS
211 help 203 help
212 If this is set, STACKLEAK metrics fo 204 If this is set, STACKLEAK metrics for every task are available in
213 the /proc file system. In particular 205 the /proc file system. In particular, /proc/<pid>/stack_depth
214 shows the maximum kernel stack consu 206 shows the maximum kernel stack consumption for the current and
215 previous syscalls. Although this inf 207 previous syscalls. Although this information is not precise, it
216 can be useful for estimating the STA 208 can be useful for estimating the STACKLEAK performance impact for
217 your workloads. 209 your workloads.
218 210
219 config STACKLEAK_RUNTIME_DISABLE 211 config STACKLEAK_RUNTIME_DISABLE
220 bool "Allow runtime disabling of kerne 212 bool "Allow runtime disabling of kernel stack erasing"
221 depends on GCC_PLUGIN_STACKLEAK 213 depends on GCC_PLUGIN_STACKLEAK
222 help 214 help
223 This option provides 'stack_erasing' 215 This option provides 'stack_erasing' sysctl, which can be used in
224 runtime to control kernel stack eras 216 runtime to control kernel stack erasing for kernels built with
225 CONFIG_GCC_PLUGIN_STACKLEAK. 217 CONFIG_GCC_PLUGIN_STACKLEAK.
226 218
227 config INIT_ON_ALLOC_DEFAULT_ON 219 config INIT_ON_ALLOC_DEFAULT_ON
228 bool "Enable heap memory zeroing on al 220 bool "Enable heap memory zeroing on allocation by default"
229 depends on !KMSAN <<
230 help 221 help
231 This has the effect of setting "init 222 This has the effect of setting "init_on_alloc=1" on the kernel
232 command line. This can be disabled w 223 command line. This can be disabled with "init_on_alloc=0".
233 When "init_on_alloc" is enabled, all 224 When "init_on_alloc" is enabled, all page allocator and slab
234 allocator memory will be zeroed when 225 allocator memory will be zeroed when allocated, eliminating
235 many kinds of "uninitialized heap me 226 many kinds of "uninitialized heap memory" flaws, especially
236 heap content exposures. The performa 227 heap content exposures. The performance impact varies by
237 workload, but most cases see <1% imp 228 workload, but most cases see <1% impact. Some synthetic
238 workloads have measured as high as 7 229 workloads have measured as high as 7%.
239 230
240 config INIT_ON_FREE_DEFAULT_ON 231 config INIT_ON_FREE_DEFAULT_ON
241 bool "Enable heap memory zeroing on fr 232 bool "Enable heap memory zeroing on free by default"
242 depends on !KMSAN <<
243 help 233 help
244 This has the effect of setting "init 234 This has the effect of setting "init_on_free=1" on the kernel
245 command line. This can be disabled w 235 command line. This can be disabled with "init_on_free=0".
246 Similar to "init_on_alloc", when "in 236 Similar to "init_on_alloc", when "init_on_free" is enabled,
247 all page allocator and slab allocato 237 all page allocator and slab allocator memory will be zeroed
248 when freed, eliminating many kinds o 238 when freed, eliminating many kinds of "uninitialized heap memory"
249 flaws, especially heap content expos 239 flaws, especially heap content exposures. The primary difference
250 with "init_on_free" is that data lif 240 with "init_on_free" is that data lifetime in memory is reduced,
251 as anything freed is wiped immediate 241 as anything freed is wiped immediately, making live forensics or
252 cold boot memory attacks unable to r 242 cold boot memory attacks unable to recover freed memory contents.
253 The performance impact varies by wor 243 The performance impact varies by workload, but is more expensive
254 than "init_on_alloc" due to the nega 244 than "init_on_alloc" due to the negative cache effects of
255 touching "cold" memory areas. Most c 245 touching "cold" memory areas. Most cases see 3-5% impact. Some
256 synthetic workloads have measured as 246 synthetic workloads have measured as high as 8%.
257 247
258 config CC_HAS_ZERO_CALL_USED_REGS 248 config CC_HAS_ZERO_CALL_USED_REGS
259 def_bool $(cc-option,-fzero-call-used- 249 def_bool $(cc-option,-fzero-call-used-regs=used-gpr)
260 # https://github.com/ClangBuiltLinux/l <<
261 # https://github.com/llvm/llvm-project <<
262 depends on !CC_IS_CLANG || CLANG_VERSI <<
263 250
264 config ZERO_CALL_USED_REGS 251 config ZERO_CALL_USED_REGS
265 bool "Enable register zeroing on funct 252 bool "Enable register zeroing on function exit"
266 depends on CC_HAS_ZERO_CALL_USED_REGS 253 depends on CC_HAS_ZERO_CALL_USED_REGS
267 help 254 help
268 At the end of functions, always zero 255 At the end of functions, always zero any caller-used register
269 contents. This helps ensure that tem 256 contents. This helps ensure that temporary values are not
270 leaked beyond the function boundary. 257 leaked beyond the function boundary. This means that register
271 contents are less likely to be avail 258 contents are less likely to be available for side channels
272 and information exposures. Additiona 259 and information exposures. Additionally, this helps reduce the
273 number of useful ROP gadgets by abou 260 number of useful ROP gadgets by about 20% (and removes compiler
274 generated "write-what-where" gadgets 261 generated "write-what-where" gadgets) in the resulting kernel
275 image. This has a less than 1% perfo 262 image. This has a less than 1% performance impact on most
276 workloads. Image size growth depends 263 workloads. Image size growth depends on architecture, and should
277 be evaluated for suitability. For ex 264 be evaluated for suitability. For example, x86_64 grows by less
278 than 1%, and arm64 grows by about 5% 265 than 1%, and arm64 grows by about 5%.
279 266
280 endmenu 267 endmenu
281 <<
282 menu "Hardening of kernel data structures" <<
283 <<
284 config LIST_HARDENED <<
285 bool "Check integrity of linked list m <<
286 help <<
287 Minimal integrity checking in the li <<
288 to catch memory corruptions that are <<
289 immediate access fault. <<
290 <<
291 If unsure, say N. <<
292 <<
293 config BUG_ON_DATA_CORRUPTION <<
294 bool "Trigger a BUG when data corrupti <<
295 select LIST_HARDENED <<
296 help <<
297 Select this option if the kernel sho <<
298 data corruption in kernel memory str <<
299 for validity. <<
300 <<
301 If unsure, say N. <<
302 <<
303 endmenu <<
304 <<
305 config CC_HAS_RANDSTRUCT <<
306 def_bool $(cc-option,-frandomize-layou <<
307 # Randstruct was first added in Clang <<
308 # Clang 16 due to https://github.com/l <<
309 depends on !CC_IS_CLANG || CLANG_VERSI <<
310 <<
311 choice <<
312 prompt "Randomize layout of sensitive <<
313 default RANDSTRUCT_FULL if COMPILE_TES <<
314 default RANDSTRUCT_NONE <<
315 help <<
316 If you enable this, the layouts of s <<
317 function pointers (and have not been <<
318 __no_randomize_layout), or structure <<
319 marked with __randomize_layout, will <<
320 This can introduce the requirement o <<
321 exposure vulnerability for exploits <<
322 types. <<
323 <<
324 Enabling this feature will introduce <<
325 slightly increase memory usage, and <<
326 tools like Volatility against the sy <<
327 source tree isn't cleaned after kern <<
328 <<
329 The seed used for compilation is in <<
330 It remains after a "make clean" to a <<
331 be compiled with the existing seed a <<
332 "make mrproper" or "make distclean". <<
333 public, or the structure layout can <<
334 <<
335 config RANDSTRUCT_NONE <<
336 bool "Disable structure layout <<
337 help <<
338 Build normally: no structure <<
339 <<
340 config RANDSTRUCT_FULL <<
341 bool "Fully randomize structur <<
342 depends on CC_HAS_RANDSTRUCT | <<
343 select MODVERSIONS if MODULES <<
344 help <<
345 Fully randomize the member l <<
346 structures as much as possib <<
347 memory size and performance <<
348 <<
349 One difference between the C <<
350 implementations is the handl <<
351 plugin treats them as fully <<
352 introducing sometimes signif <<
353 to keep adjacent bitfields t <<
354 ordering randomized. <<
355 <<
356 config RANDSTRUCT_PERFORMANCE <<
357 bool "Limit randomization of s <<
358 depends on GCC_PLUGINS <<
359 select MODVERSIONS if MODULES <<
360 help <<
361 Randomization of sensitive k <<
362 best effort at restricting r <<
363 groups of members. It will f <<
364 in structures. This reduces <<
365 at the cost of weakened rand <<
366 endchoice <<
367 <<
368 config RANDSTRUCT <<
369 def_bool !RANDSTRUCT_NONE <<
370 <<
371 config GCC_PLUGIN_RANDSTRUCT <<
372 def_bool GCC_PLUGINS && RANDSTRUCT <<
373 help <<
374 Use GCC plugin to randomize structur <<
375 <<
376 This plugin was ported from grsecuri <<
377 information at: <<
378 * https://grsecurity.net/ <<
379 * https://pax.grsecurity.net/ <<
380 268
381 endmenu 269 endmenu
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.