Loading...
1.. SPDX-License-Identifier: GPL-2.0
2.. Copyright (C) 2023, Google LLC.
3
4Kernel Address Sanitizer (KASAN)
5================================
6
7Overview
8--------
9
10Kernel Address Sanitizer (KASAN) is a dynamic memory safety error detector
11designed to find out-of-bounds and use-after-free bugs.
12
13KASAN has three modes:
14
151. Generic KASAN
162. Software Tag-Based KASAN
173. Hardware Tag-Based KASAN
18
19Generic KASAN, enabled with CONFIG_KASAN_GENERIC, is the mode intended for
20debugging, similar to userspace ASan. This mode is supported on many CPU
21architectures, but it has significant performance and memory overheads.
22
23Software Tag-Based KASAN or SW_TAGS KASAN, enabled with CONFIG_KASAN_SW_TAGS,
24can be used for both debugging and dogfood testing, similar to userspace HWASan.
25This mode is only supported for arm64, but its moderate memory overhead allows
26using it for testing on memory-restricted devices with real workloads.
27
28Hardware Tag-Based KASAN or HW_TAGS KASAN, enabled with CONFIG_KASAN_HW_TAGS,
29is the mode intended to be used as an in-field memory bug detector or as a
30security mitigation. This mode only works on arm64 CPUs that support MTE
31(Memory Tagging Extension), but it has low memory and performance overheads and
32thus can be used in production.
33
34For details about the memory and performance impact of each KASAN mode, see the
35descriptions of the corresponding Kconfig options.
36
37The Generic and the Software Tag-Based modes are commonly referred to as the
38software modes. The Software Tag-Based and the Hardware Tag-Based modes are
39referred to as the tag-based modes.
40
41Support
42-------
43
44Architectures
45~~~~~~~~~~~~~
46
47Generic KASAN is supported on x86_64, arm, arm64, powerpc, riscv, s390, xtensa,
48and loongarch, and the tag-based KASAN modes are supported only on arm64.
49
50Compilers
51~~~~~~~~~
52
53Software KASAN modes use compile-time instrumentation to insert validity checks
54before every memory access and thus require a compiler version that provides
55support for that. The Hardware Tag-Based mode relies on hardware to perform
56these checks but still requires a compiler version that supports the memory
57tagging instructions.
58
59Generic KASAN requires GCC version 8.3.0 or later
60or any Clang version supported by the kernel.
61
62Software Tag-Based KASAN requires GCC 11+
63or any Clang version supported by the kernel.
64
65Hardware Tag-Based KASAN requires GCC 10+ or Clang 12+.
66
67Memory types
68~~~~~~~~~~~~
69
70Generic KASAN supports finding bugs in all of slab, page_alloc, vmap, vmalloc,
71stack, and global memory.
72
73Software Tag-Based KASAN supports slab, page_alloc, vmalloc, and stack memory.
74
75Hardware Tag-Based KASAN supports slab, page_alloc, and non-executable vmalloc
76memory.
77
78For slab, both software KASAN modes support SLUB and SLAB allocators, while
79Hardware Tag-Based KASAN only supports SLUB.
80
81Usage
82-----
83
84To enable KASAN, configure the kernel with::
85
86 CONFIG_KASAN=y
87
88and choose between ``CONFIG_KASAN_GENERIC`` (to enable Generic KASAN),
89``CONFIG_KASAN_SW_TAGS`` (to enable Software Tag-Based KASAN), and
90``CONFIG_KASAN_HW_TAGS`` (to enable Hardware Tag-Based KASAN).
91
92For the software modes, also choose between ``CONFIG_KASAN_OUTLINE`` and
93``CONFIG_KASAN_INLINE``. Outline and inline are compiler instrumentation types.
94The former produces a smaller binary while the latter is up to 2 times faster.
95
96To include alloc and free stack traces of affected slab objects into reports,
97enable ``CONFIG_STACKTRACE``. To include alloc and free stack traces of affected
98physical pages, enable ``CONFIG_PAGE_OWNER`` and boot with ``page_owner=on``.
99
100Boot parameters
101~~~~~~~~~~~~~~~
102
103KASAN is affected by the generic ``panic_on_warn`` command line parameter.
104When it is enabled, KASAN panics the kernel after printing a bug report.
105
106By default, KASAN prints a bug report only for the first invalid memory access.
107With ``kasan_multi_shot``, KASAN prints a report on every invalid access. This
108effectively disables ``panic_on_warn`` for KASAN reports.
109
110Alternatively, independent of ``panic_on_warn``, the ``kasan.fault=`` boot
111parameter can be used to control panic and reporting behaviour:
112
113- ``kasan.fault=report``, ``=panic``, or ``=panic_on_write`` controls whether
114 to only print a KASAN report, panic the kernel, or panic the kernel on
115 invalid writes only (default: ``report``). The panic happens even if
116 ``kasan_multi_shot`` is enabled. Note that when using asynchronous mode of
117 Hardware Tag-Based KASAN, ``kasan.fault=panic_on_write`` always panics on
118 asynchronously checked accesses (including reads).
119
120Software and Hardware Tag-Based KASAN modes (see the section about various
121modes below) support altering stack trace collection behavior:
122
123- ``kasan.stacktrace=off`` or ``=on`` disables or enables alloc and free stack
124 traces collection (default: ``on``).
125- ``kasan.stack_ring_size=<number of entries>`` specifies the number of entries
126 in the stack ring (default: ``32768``).
127
128Hardware Tag-Based KASAN mode is intended for use in production as a security
129mitigation. Therefore, it supports additional boot parameters that allow
130disabling KASAN altogether or controlling its features:
131
132- ``kasan=off`` or ``=on`` controls whether KASAN is enabled (default: ``on``).
133
134- ``kasan.mode=sync``, ``=async`` or ``=asymm`` controls whether KASAN
135 is configured in synchronous, asynchronous or asymmetric mode of
136 execution (default: ``sync``).
137 Synchronous mode: a bad access is detected immediately when a tag
138 check fault occurs.
139 Asynchronous mode: a bad access detection is delayed. When a tag check
140 fault occurs, the information is stored in hardware (in the TFSR_EL1
141 register for arm64). The kernel periodically checks the hardware and
142 only reports tag faults during these checks.
143 Asymmetric mode: a bad access is detected synchronously on reads and
144 asynchronously on writes.
145
146- ``kasan.vmalloc=off`` or ``=on`` disables or enables tagging of vmalloc
147 allocations (default: ``on``).
148
149- ``kasan.page_alloc.sample=<sampling interval>`` makes KASAN tag only every
150 Nth page_alloc allocation with the order equal or greater than
151 ``kasan.page_alloc.sample.order``, where N is the value of the ``sample``
152 parameter (default: ``1``, or tag every such allocation).
153 This parameter is intended to mitigate the performance overhead introduced
154 by KASAN.
155 Note that enabling this parameter makes Hardware Tag-Based KASAN skip checks
156 of allocations chosen by sampling and thus miss bad accesses to these
157 allocations. Use the default value for accurate bug detection.
158
159- ``kasan.page_alloc.sample.order=<minimum page order>`` specifies the minimum
160 order of allocations that are affected by sampling (default: ``3``).
161 Only applies when ``kasan.page_alloc.sample`` is set to a value greater
162 than ``1``.
163 This parameter is intended to allow sampling only large page_alloc
164 allocations, which is the biggest source of the performance overhead.
165
166Error reports
167~~~~~~~~~~~~~
168
169A typical KASAN report looks like this::
170
171 ==================================================================
172 BUG: KASAN: slab-out-of-bounds in kmalloc_oob_right+0xa8/0xbc [kasan_test]
173 Write of size 1 at addr ffff8801f44ec37b by task insmod/2760
174
175 CPU: 1 PID: 2760 Comm: insmod Not tainted 4.19.0-rc3+ #698
176 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1 04/01/2014
177 Call Trace:
178 dump_stack+0x94/0xd8
179 print_address_description+0x73/0x280
180 kasan_report+0x144/0x187
181 __asan_report_store1_noabort+0x17/0x20
182 kmalloc_oob_right+0xa8/0xbc [kasan_test]
183 kmalloc_tests_init+0x16/0x700 [kasan_test]
184 do_one_initcall+0xa5/0x3ae
185 do_init_module+0x1b6/0x547
186 load_module+0x75df/0x8070
187 __do_sys_init_module+0x1c6/0x200
188 __x64_sys_init_module+0x6e/0xb0
189 do_syscall_64+0x9f/0x2c0
190 entry_SYSCALL_64_after_hwframe+0x44/0xa9
191 RIP: 0033:0x7f96443109da
192 RSP: 002b:00007ffcf0b51b08 EFLAGS: 00000202 ORIG_RAX: 00000000000000af
193 RAX: ffffffffffffffda RBX: 000055dc3ee521a0 RCX: 00007f96443109da
194 RDX: 00007f96445cff88 RSI: 0000000000057a50 RDI: 00007f9644992000
195 RBP: 000055dc3ee510b0 R08: 0000000000000003 R09: 0000000000000000
196 R10: 00007f964430cd0a R11: 0000000000000202 R12: 00007f96445cff88
197 R13: 000055dc3ee51090 R14: 0000000000000000 R15: 0000000000000000
198
199 Allocated by task 2760:
200 save_stack+0x43/0xd0
201 kasan_kmalloc+0xa7/0xd0
202 kmem_cache_alloc_trace+0xe1/0x1b0
203 kmalloc_oob_right+0x56/0xbc [kasan_test]
204 kmalloc_tests_init+0x16/0x700 [kasan_test]
205 do_one_initcall+0xa5/0x3ae
206 do_init_module+0x1b6/0x547
207 load_module+0x75df/0x8070
208 __do_sys_init_module+0x1c6/0x200
209 __x64_sys_init_module+0x6e/0xb0
210 do_syscall_64+0x9f/0x2c0
211 entry_SYSCALL_64_after_hwframe+0x44/0xa9
212
213 Freed by task 815:
214 save_stack+0x43/0xd0
215 __kasan_slab_free+0x135/0x190
216 kasan_slab_free+0xe/0x10
217 kfree+0x93/0x1a0
218 umh_complete+0x6a/0xa0
219 call_usermodehelper_exec_async+0x4c3/0x640
220 ret_from_fork+0x35/0x40
221
222 The buggy address belongs to the object at ffff8801f44ec300
223 which belongs to the cache kmalloc-128 of size 128
224 The buggy address is located 123 bytes inside of
225 128-byte region [ffff8801f44ec300, ffff8801f44ec380)
226 The buggy address belongs to the page:
227 page:ffffea0007d13b00 count:1 mapcount:0 mapping:ffff8801f7001640 index:0x0
228 flags: 0x200000000000100(slab)
229 raw: 0200000000000100 ffffea0007d11dc0 0000001a0000001a ffff8801f7001640
230 raw: 0000000000000000 0000000080150015 00000001ffffffff 0000000000000000
231 page dumped because: kasan: bad access detected
232
233 Memory state around the buggy address:
234 ffff8801f44ec200: fc fc fc fc fc fc fc fc fb fb fb fb fb fb fb fb
235 ffff8801f44ec280: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
236 >ffff8801f44ec300: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 03
237 ^
238 ffff8801f44ec380: fc fc fc fc fc fc fc fc fb fb fb fb fb fb fb fb
239 ffff8801f44ec400: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
240 ==================================================================
241
242The report header summarizes what kind of bug happened and what kind of access
243caused it. It is followed by a stack trace of the bad access, a stack trace of
244where the accessed memory was allocated (in case a slab object was accessed),
245and a stack trace of where the object was freed (in case of a use-after-free
246bug report). Next comes a description of the accessed slab object and the
247information about the accessed memory page.
248
249In the end, the report shows the memory state around the accessed address.
250Internally, KASAN tracks memory state separately for each memory granule, which
251is either 8 or 16 aligned bytes depending on KASAN mode. Each number in the
252memory state section of the report shows the state of one of the memory
253granules that surround the accessed address.
254
255For Generic KASAN, the size of each memory granule is 8. The state of each
256granule is encoded in one shadow byte. Those 8 bytes can be accessible,
257partially accessible, freed, or be a part of a redzone. KASAN uses the following
258encoding for each shadow byte: 00 means that all 8 bytes of the corresponding
259memory region are accessible; number N (1 <= N <= 7) means that the first N
260bytes are accessible, and other (8 - N) bytes are not; any negative value
261indicates that the entire 8-byte word is inaccessible. KASAN uses different
262negative values to distinguish between different kinds of inaccessible memory
263like redzones or freed memory (see mm/kasan/kasan.h).
264
265In the report above, the arrow points to the shadow byte ``03``, which means
266that the accessed address is partially accessible.
267
268For tag-based KASAN modes, this last report section shows the memory tags around
269the accessed address (see the `Implementation details`_ section).
270
271Note that KASAN bug titles (like ``slab-out-of-bounds`` or ``use-after-free``)
272are best-effort: KASAN prints the most probable bug type based on the limited
273information it has. The actual type of the bug might be different.
274
275Generic KASAN also reports up to two auxiliary call stack traces. These stack
276traces point to places in code that interacted with the object but that are not
277directly present in the bad access stack trace. Currently, this includes
278call_rcu() and workqueue queuing.
279
280CONFIG_KASAN_EXTRA_INFO
281~~~~~~~~~~~~~~~~~~~~~~~
282
283Enabling CONFIG_KASAN_EXTRA_INFO allows KASAN to record and report more
284information. The extra information currently supported is the CPU number and
285timestamp at allocation and free. More information can help find the cause of
286the bug and correlate the error with other system events, at the cost of using
287extra memory to record more information (more cost details in the help text of
288CONFIG_KASAN_EXTRA_INFO).
289
290Here is the report with CONFIG_KASAN_EXTRA_INFO enabled (only the
291different parts are shown)::
292
293 ==================================================================
294 ...
295 Allocated by task 134 on cpu 5 at 229.133855s:
296 ...
297 Freed by task 136 on cpu 3 at 230.199335s:
298 ...
299 ==================================================================
300
301Implementation details
302----------------------
303
304Generic KASAN
305~~~~~~~~~~~~~
306
307Software KASAN modes use shadow memory to record whether each byte of memory is
308safe to access and use compile-time instrumentation to insert shadow memory
309checks before each memory access.
310
311Generic KASAN dedicates 1/8th of kernel memory to its shadow memory (16TB
312to cover 128TB on x86_64) and uses direct mapping with a scale and offset to
313translate a memory address to its corresponding shadow address.
314
315Here is the function which translates an address to its corresponding shadow
316address::
317
318 static inline void *kasan_mem_to_shadow(const void *addr)
319 {
320 return (void *)((unsigned long)addr >> KASAN_SHADOW_SCALE_SHIFT)
321 + KASAN_SHADOW_OFFSET;
322 }
323
324where ``KASAN_SHADOW_SCALE_SHIFT = 3``.
325
326Compile-time instrumentation is used to insert memory access checks. Compiler
327inserts function calls (``__asan_load*(addr)``, ``__asan_store*(addr)``) before
328each memory access of size 1, 2, 4, 8, or 16. These functions check whether
329memory accesses are valid or not by checking corresponding shadow memory.
330
331With inline instrumentation, instead of making function calls, the compiler
332directly inserts the code to check shadow memory. This option significantly
333enlarges the kernel, but it gives an x1.1-x2 performance boost over the
334outline-instrumented kernel.
335
336Generic KASAN is the only mode that delays the reuse of freed objects via
337quarantine (see mm/kasan/quarantine.c for implementation).
338
339Software Tag-Based KASAN
340~~~~~~~~~~~~~~~~~~~~~~~~
341
342Software Tag-Based KASAN uses a software memory tagging approach to checking
343access validity. It is currently only implemented for the arm64 architecture.
344
345Software Tag-Based KASAN uses the Top Byte Ignore (TBI) feature of arm64 CPUs
346to store a pointer tag in the top byte of kernel pointers. It uses shadow memory
347to store memory tags associated with each 16-byte memory cell (therefore, it
348dedicates 1/16th of the kernel memory for shadow memory).
349
350On each memory allocation, Software Tag-Based KASAN generates a random tag, tags
351the allocated memory with this tag, and embeds the same tag into the returned
352pointer.
353
354Software Tag-Based KASAN uses compile-time instrumentation to insert checks
355before each memory access. These checks make sure that the tag of the memory
356that is being accessed is equal to the tag of the pointer that is used to access
357this memory. In case of a tag mismatch, Software Tag-Based KASAN prints a bug
358report.
359
360Software Tag-Based KASAN also has two instrumentation modes (outline, which
361emits callbacks to check memory accesses; and inline, which performs the shadow
362memory checks inline). With outline instrumentation mode, a bug report is
363printed from the function that performs the access check. With inline
364instrumentation, a ``brk`` instruction is emitted by the compiler, and a
365dedicated ``brk`` handler is used to print bug reports.
366
367Software Tag-Based KASAN uses 0xFF as a match-all pointer tag (accesses through
368pointers with the 0xFF pointer tag are not checked). The value 0xFE is currently
369reserved to tag freed memory regions.
370
371Hardware Tag-Based KASAN
372~~~~~~~~~~~~~~~~~~~~~~~~
373
374Hardware Tag-Based KASAN is similar to the software mode in concept but uses
375hardware memory tagging support instead of compiler instrumentation and
376shadow memory.
377
378Hardware Tag-Based KASAN is currently only implemented for arm64 architecture
379and based on both arm64 Memory Tagging Extension (MTE) introduced in ARMv8.5
380Instruction Set Architecture and Top Byte Ignore (TBI).
381
382Special arm64 instructions are used to assign memory tags for each allocation.
383Same tags are assigned to pointers to those allocations. On every memory
384access, hardware makes sure that the tag of the memory that is being accessed is
385equal to the tag of the pointer that is used to access this memory. In case of a
386tag mismatch, a fault is generated, and a report is printed.
387
388Hardware Tag-Based KASAN uses 0xFF as a match-all pointer tag (accesses through
389pointers with the 0xFF pointer tag are not checked). The value 0xFE is currently
390reserved to tag freed memory regions.
391
392If the hardware does not support MTE (pre ARMv8.5), Hardware Tag-Based KASAN
393will not be enabled. In this case, all KASAN boot parameters are ignored.
394
395Note that enabling CONFIG_KASAN_HW_TAGS always results in in-kernel TBI being
396enabled. Even when ``kasan.mode=off`` is provided or when the hardware does not
397support MTE (but supports TBI).
398
399Hardware Tag-Based KASAN only reports the first found bug. After that, MTE tag
400checking gets disabled.
401
402Shadow memory
403-------------
404
405The contents of this section are only applicable to software KASAN modes.
406
407The kernel maps memory in several different parts of the address space.
408The range of kernel virtual addresses is large: there is not enough real
409memory to support a real shadow region for every address that could be
410accessed by the kernel. Therefore, KASAN only maps real shadow for certain
411parts of the address space.
412
413Default behaviour
414~~~~~~~~~~~~~~~~~
415
416By default, architectures only map real memory over the shadow region
417for the linear mapping (and potentially other small areas). For all
418other areas - such as vmalloc and vmemmap space - a single read-only
419page is mapped over the shadow area. This read-only shadow page
420declares all memory accesses as permitted.
421
422This presents a problem for modules: they do not live in the linear
423mapping but in a dedicated module space. By hooking into the module
424allocator, KASAN temporarily maps real shadow memory to cover them.
425This allows detection of invalid accesses to module globals, for example.
426
427This also creates an incompatibility with ``VMAP_STACK``: if the stack
428lives in vmalloc space, it will be shadowed by the read-only page, and
429the kernel will fault when trying to set up the shadow data for stack
430variables.
431
432CONFIG_KASAN_VMALLOC
433~~~~~~~~~~~~~~~~~~~~
434
435With ``CONFIG_KASAN_VMALLOC``, KASAN can cover vmalloc space at the
436cost of greater memory usage. Currently, this is supported on x86,
437arm64, riscv, s390, and powerpc.
438
439This works by hooking into vmalloc and vmap and dynamically
440allocating real shadow memory to back the mappings.
441
442Most mappings in vmalloc space are small, requiring less than a full
443page of shadow space. Allocating a full shadow page per mapping would
444therefore be wasteful. Furthermore, to ensure that different mappings
445use different shadow pages, mappings would have to be aligned to
446``KASAN_GRANULE_SIZE * PAGE_SIZE``.
447
448Instead, KASAN shares backing space across multiple mappings. It allocates
449a backing page when a mapping in vmalloc space uses a particular page
450of the shadow region. This page can be shared by other vmalloc
451mappings later on.
452
453KASAN hooks into the vmap infrastructure to lazily clean up unused shadow
454memory.
455
456To avoid the difficulties around swapping mappings around, KASAN expects
457that the part of the shadow region that covers the vmalloc space will
458not be covered by the early shadow page but will be left unmapped.
459This will require changes in arch-specific code.
460
461This allows ``VMAP_STACK`` support on x86 and can simplify support of
462architectures that do not have a fixed module region.
463
464For developers
465--------------
466
467Ignoring accesses
468~~~~~~~~~~~~~~~~~
469
470Software KASAN modes use compiler instrumentation to insert validity checks.
471Such instrumentation might be incompatible with some parts of the kernel, and
472therefore needs to be disabled.
473
474Other parts of the kernel might access metadata for allocated objects.
475Normally, KASAN detects and reports such accesses, but in some cases (e.g.,
476in memory allocators), these accesses are valid.
477
478For software KASAN modes, to disable instrumentation for a specific file or
479directory, add a ``KASAN_SANITIZE`` annotation to the respective kernel
480Makefile:
481
482- For a single file (e.g., main.o)::
483
484 KASAN_SANITIZE_main.o := n
485
486- For all files in one directory::
487
488 KASAN_SANITIZE := n
489
490For software KASAN modes, to disable instrumentation on a per-function basis,
491use the KASAN-specific ``__no_sanitize_address`` function attribute or the
492generic ``noinstr`` one.
493
494Note that disabling compiler instrumentation (either on a per-file or a
495per-function basis) makes KASAN ignore the accesses that happen directly in
496that code for software KASAN modes. It does not help when the accesses happen
497indirectly (through calls to instrumented functions) or with Hardware
498Tag-Based KASAN, which does not use compiler instrumentation.
499
500For software KASAN modes, to disable KASAN reports in a part of the kernel code
501for the current task, annotate this part of the code with a
502``kasan_disable_current()``/``kasan_enable_current()`` section. This also
503disables the reports for indirect accesses that happen through function calls.
504
505For tag-based KASAN modes, to disable access checking, use
506``kasan_reset_tag()`` or ``page_kasan_tag_reset()``. Note that temporarily
507disabling access checking via ``page_kasan_tag_reset()`` requires saving and
508restoring the per-page KASAN tag via ``page_kasan_tag``/``page_kasan_tag_set``.
509
510Tests
511~~~~~
512
513There are KASAN tests that allow verifying that KASAN works and can detect
514certain types of memory corruptions.
515
516All KASAN tests are integrated with the KUnit Test Framework and can be enabled
517via ``CONFIG_KASAN_KUNIT_TEST``. The tests can be run and partially verified
518automatically in a few different ways; see the instructions below.
519
520Each KASAN test prints one of multiple KASAN reports if an error is detected.
521Then the test prints its number and status.
522
523When a test passes::
524
525 ok 28 - kmalloc_double_kzfree
526
527When a test fails due to a failed ``kmalloc``::
528
529 # kmalloc_large_oob_right: ASSERTION FAILED at mm/kasan/kasan_test.c:245
530 Expected ptr is not null, but is
531 not ok 5 - kmalloc_large_oob_right
532
533When a test fails due to a missing KASAN report::
534
535 # kmalloc_double_kzfree: EXPECTATION FAILED at mm/kasan/kasan_test.c:709
536 KASAN failure expected in "kfree_sensitive(ptr)", but none occurred
537 not ok 28 - kmalloc_double_kzfree
538
539
540At the end the cumulative status of all KASAN tests is printed. On success::
541
542 ok 1 - kasan
543
544Or, if one of the tests failed::
545
546 not ok 1 - kasan
547
548There are a few ways to run the KASAN tests.
549
5501. Loadable module
551
552 With ``CONFIG_KUNIT`` enabled, the tests can be built as a loadable module
553 and run by loading ``kasan_test.ko`` with ``insmod`` or ``modprobe``.
554
5552. Built-In
556
557 With ``CONFIG_KUNIT`` built-in, the tests can be built-in as well.
558 In this case, the tests will run at boot as a late-init call.
559
5603. Using kunit_tool
561
562 With ``CONFIG_KUNIT`` and ``CONFIG_KASAN_KUNIT_TEST`` built-in, it is also
563 possible to use ``kunit_tool`` to see the results of KUnit tests in a more
564 readable way. This will not print the KASAN reports of the tests that passed.
565 See `KUnit documentation <https://www.kernel.org/doc/html/latest/dev-tools/kunit/index.html>`_
566 for more up-to-date information on ``kunit_tool``.
567
568.. _KUnit: https://www.kernel.org/doc/html/latest/dev-tools/kunit/index.html
1The Kernel Address Sanitizer (KASAN)
2====================================
3
4Overview
5--------
6
7KernelAddressSANitizer (KASAN) is a dynamic memory error detector. It provides
8a fast and comprehensive solution for finding use-after-free and out-of-bounds
9bugs.
10
11KASAN uses compile-time instrumentation for checking every memory access,
12therefore you will need a GCC version 4.9.2 or later. GCC 5.0 or later is
13required for detection of out-of-bounds accesses to stack or global variables.
14
15Currently KASAN is supported only for the x86_64 and arm64 architectures.
16
17Usage
18-----
19
20To enable KASAN configure kernel with::
21
22 CONFIG_KASAN = y
23
24and choose between CONFIG_KASAN_OUTLINE and CONFIG_KASAN_INLINE. Outline and
25inline are compiler instrumentation types. The former produces smaller binary
26the latter is 1.1 - 2 times faster. Inline instrumentation requires a GCC
27version 5.0 or later.
28
29KASAN works with both SLUB and SLAB memory allocators.
30For better bug detection and nicer reporting, enable CONFIG_STACKTRACE.
31
32To disable instrumentation for specific files or directories, add a line
33similar to the following to the respective kernel Makefile:
34
35- For a single file (e.g. main.o)::
36
37 KASAN_SANITIZE_main.o := n
38
39- For all files in one directory::
40
41 KASAN_SANITIZE := n
42
43Error reports
44~~~~~~~~~~~~~
45
46A typical out of bounds access report looks like this::
47
48 ==================================================================
49 BUG: AddressSanitizer: out of bounds access in kmalloc_oob_right+0x65/0x75 [test_kasan] at addr ffff8800693bc5d3
50 Write of size 1 by task modprobe/1689
51 =============================================================================
52 BUG kmalloc-128 (Not tainted): kasan error
53 -----------------------------------------------------------------------------
54
55 Disabling lock debugging due to kernel taint
56 INFO: Allocated in kmalloc_oob_right+0x3d/0x75 [test_kasan] age=0 cpu=0 pid=1689
57 __slab_alloc+0x4b4/0x4f0
58 kmem_cache_alloc_trace+0x10b/0x190
59 kmalloc_oob_right+0x3d/0x75 [test_kasan]
60 init_module+0x9/0x47 [test_kasan]
61 do_one_initcall+0x99/0x200
62 load_module+0x2cb3/0x3b20
63 SyS_finit_module+0x76/0x80
64 system_call_fastpath+0x12/0x17
65 INFO: Slab 0xffffea0001a4ef00 objects=17 used=7 fp=0xffff8800693bd728 flags=0x100000000004080
66 INFO: Object 0xffff8800693bc558 @offset=1368 fp=0xffff8800693bc720
67
68 Bytes b4 ffff8800693bc548: 00 00 00 00 00 00 00 00 5a 5a 5a 5a 5a 5a 5a 5a ........ZZZZZZZZ
69 Object ffff8800693bc558: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk
70 Object ffff8800693bc568: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk
71 Object ffff8800693bc578: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk
72 Object ffff8800693bc588: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk
73 Object ffff8800693bc598: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk
74 Object ffff8800693bc5a8: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk
75 Object ffff8800693bc5b8: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk
76 Object ffff8800693bc5c8: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b a5 kkkkkkkkkkkkkkk.
77 Redzone ffff8800693bc5d8: cc cc cc cc cc cc cc cc ........
78 Padding ffff8800693bc718: 5a 5a 5a 5a 5a 5a 5a 5a ZZZZZZZZ
79 CPU: 0 PID: 1689 Comm: modprobe Tainted: G B 3.18.0-rc1-mm1+ #98
80 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.7.5-0-ge51488c-20140602_164612-nilsson.home.kraxel.org 04/01/2014
81 ffff8800693bc000 0000000000000000 ffff8800693bc558 ffff88006923bb78
82 ffffffff81cc68ae 00000000000000f3 ffff88006d407600 ffff88006923bba8
83 ffffffff811fd848 ffff88006d407600 ffffea0001a4ef00 ffff8800693bc558
84 Call Trace:
85 [<ffffffff81cc68ae>] dump_stack+0x46/0x58
86 [<ffffffff811fd848>] print_trailer+0xf8/0x160
87 [<ffffffffa00026a7>] ? kmem_cache_oob+0xc3/0xc3 [test_kasan]
88 [<ffffffff811ff0f5>] object_err+0x35/0x40
89 [<ffffffffa0002065>] ? kmalloc_oob_right+0x65/0x75 [test_kasan]
90 [<ffffffff8120b9fa>] kasan_report_error+0x38a/0x3f0
91 [<ffffffff8120a79f>] ? kasan_poison_shadow+0x2f/0x40
92 [<ffffffff8120b344>] ? kasan_unpoison_shadow+0x14/0x40
93 [<ffffffff8120a79f>] ? kasan_poison_shadow+0x2f/0x40
94 [<ffffffffa00026a7>] ? kmem_cache_oob+0xc3/0xc3 [test_kasan]
95 [<ffffffff8120a995>] __asan_store1+0x75/0xb0
96 [<ffffffffa0002601>] ? kmem_cache_oob+0x1d/0xc3 [test_kasan]
97 [<ffffffffa0002065>] ? kmalloc_oob_right+0x65/0x75 [test_kasan]
98 [<ffffffffa0002065>] kmalloc_oob_right+0x65/0x75 [test_kasan]
99 [<ffffffffa00026b0>] init_module+0x9/0x47 [test_kasan]
100 [<ffffffff810002d9>] do_one_initcall+0x99/0x200
101 [<ffffffff811e4e5c>] ? __vunmap+0xec/0x160
102 [<ffffffff81114f63>] load_module+0x2cb3/0x3b20
103 [<ffffffff8110fd70>] ? m_show+0x240/0x240
104 [<ffffffff81115f06>] SyS_finit_module+0x76/0x80
105 [<ffffffff81cd3129>] system_call_fastpath+0x12/0x17
106 Memory state around the buggy address:
107 ffff8800693bc300: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
108 ffff8800693bc380: fc fc 00 00 00 00 00 00 00 00 00 00 00 00 00 fc
109 ffff8800693bc400: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
110 ffff8800693bc480: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
111 ffff8800693bc500: fc fc fc fc fc fc fc fc fc fc fc 00 00 00 00 00
112 >ffff8800693bc580: 00 00 00 00 00 00 00 00 00 00 03 fc fc fc fc fc
113 ^
114 ffff8800693bc600: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
115 ffff8800693bc680: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
116 ffff8800693bc700: fc fc fc fc fb fb fb fb fb fb fb fb fb fb fb fb
117 ffff8800693bc780: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
118 ffff8800693bc800: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
119 ==================================================================
120
121The header of the report discribe what kind of bug happened and what kind of
122access caused it. It's followed by the description of the accessed slub object
123(see 'SLUB Debug output' section in Documentation/vm/slub.txt for details) and
124the description of the accessed memory page.
125
126In the last section the report shows memory state around the accessed address.
127Reading this part requires some understanding of how KASAN works.
128
129The state of each 8 aligned bytes of memory is encoded in one shadow byte.
130Those 8 bytes can be accessible, partially accessible, freed or be a redzone.
131We use the following encoding for each shadow byte: 0 means that all 8 bytes
132of the corresponding memory region are accessible; number N (1 <= N <= 7) means
133that the first N bytes are accessible, and other (8 - N) bytes are not;
134any negative value indicates that the entire 8-byte word is inaccessible.
135We use different negative values to distinguish between different kinds of
136inaccessible memory like redzones or freed memory (see mm/kasan/kasan.h).
137
138In the report above the arrows point to the shadow byte 03, which means that
139the accessed address is partially accessible.
140
141
142Implementation details
143----------------------
144
145From a high level, our approach to memory error detection is similar to that
146of kmemcheck: use shadow memory to record whether each byte of memory is safe
147to access, and use compile-time instrumentation to check shadow memory on each
148memory access.
149
150AddressSanitizer dedicates 1/8 of kernel memory to its shadow memory
151(e.g. 16TB to cover 128TB on x86_64) and uses direct mapping with a scale and
152offset to translate a memory address to its corresponding shadow address.
153
154Here is the function which translates an address to its corresponding shadow
155address::
156
157 static inline void *kasan_mem_to_shadow(const void *addr)
158 {
159 return ((unsigned long)addr >> KASAN_SHADOW_SCALE_SHIFT)
160 + KASAN_SHADOW_OFFSET;
161 }
162
163where ``KASAN_SHADOW_SCALE_SHIFT = 3``.
164
165Compile-time instrumentation used for checking memory accesses. Compiler inserts
166function calls (__asan_load*(addr), __asan_store*(addr)) before each memory
167access of size 1, 2, 4, 8 or 16. These functions check whether memory access is
168valid or not by checking corresponding shadow memory.
169
170GCC 5.0 has possibility to perform inline instrumentation. Instead of making
171function calls GCC directly inserts the code to check the shadow memory.
172This option significantly enlarges kernel but it gives x1.1-x2 performance
173boost over outline instrumented kernel.