summaryrefslogtreecommitdiff
path: root/tools/objtool/Documentation/stack-validation.txt
blob: 30f38fdc0d56cb769b589f4bd4de452758344964 (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
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
Compile-time stack metadata validation
======================================


Overview
--------

The kernel CONFIG_STACK_VALIDATION option enables a host tool named
objtool which runs at compile time.  It has a "check" subcommand which
analyzes every .o file and ensures the validity of its stack metadata.
It enforces a set of rules on asm code and C inline assembly code so
that stack traces can be reliable.

For each function, it recursively follows all possible code paths and
validates the correct frame pointer state at each instruction.

It also follows code paths involving special sections, like
.altinstructions, __jump_table, and __ex_table, which can add
alternative execution paths to a given instruction (or set of
instructions).  Similarly, it knows how to follow switch statements, for
which gcc sometimes uses jump tables.

(Objtool also has an 'orc generate' subcommand which generates debuginfo
for the ORC unwinder.  See Documentation/x86/orc-unwinder.rst in the
kernel tree for more details.)


Why do we need stack metadata validation?
-----------------------------------------

Here are some of the benefits of validating stack metadata:

a) More reliable stack traces for frame pointer enabled kernels

   Frame pointers are used for debugging purposes.  They allow runtime
   code and debug tools to be able to walk the stack to determine the
   chain of function call sites that led to the currently executing
   code.

   For some architectures, frame pointers are enabled by
   CONFIG_FRAME_POINTER.  For some other architectures they may be
   required by the ABI (sometimes referred to as "backchain pointers").

   For C code, gcc automatically generates instructions for setting up
   frame pointers when the -fno-omit-frame-pointer option is used.

   But for asm code, the frame setup instructions have to be written by
   hand, which most people don't do.  So the end result is that
   CONFIG_FRAME_POINTER is honored for C code but not for most asm code.

   For stack traces based on frame pointers to be reliable, all
   functions which call other functions must first create a stack frame
   and update the frame pointer.  If a first function doesn't properly
   create a stack frame before calling a second function, the *caller*
   of the first function will be skipped on the stack trace.

   For example, consider the following example backtrace with frame
   pointers enabled:

     [<ffffffff81812584>] dump_stack+0x4b/0x63
     [<ffffffff812d6dc2>] cmdline_proc_show+0x12/0x30
     [<ffffffff8127f568>] seq_read+0x108/0x3e0
     [<ffffffff812cce62>] proc_reg_read+0x42/0x70
     [<ffffffff81256197>] __vfs_read+0x37/0x100
     [<ffffffff81256b16>] vfs_read+0x86/0x130
     [<ffffffff81257898>] SyS_read+0x58/0xd0
     [<ffffffff8181c1f2>] entry_SYSCALL_64_fastpath+0x12/0x76

   It correctly shows that the caller of cmdline_proc_show() is
   seq_read().

   If we remove the frame pointer logic from cmdline_proc_show() by
   replacing the frame pointer related instructions with nops, here's
   what it looks like instead:

     [<ffffffff81812584>] dump_stack+0x4b/0x63
     [<ffffffff812d6dc2>] cmdline_proc_show+0x12/0x30
     [<ffffffff812cce62>] proc_reg_read+0x42/0x70
     [<ffffffff81256197>] __vfs_read+0x37/0x100
     [<ffffffff81256b16>] vfs_read+0x86/0x130
     [<ffffffff81257898>] SyS_read+0x58/0xd0
     [<ffffffff8181c1f2>] entry_SYSCALL_64_fastpath+0x12/0x76

   Notice that cmdline_proc_show()'s caller, seq_read(), has been
   skipped.  Instead the stack trace seems to show that
   cmdline_proc_show() was called by proc_reg_read().

   The benefit of objtool here is that because it ensures that *all*
   functions honor CONFIG_FRAME_POINTER, no functions will ever[*] be
   skipped on a stack trace.

   [*] unless an interrupt or exception has occurred at the very
       beginning of a function before the stack frame has been created,
       or at the very end of the function after the stack frame has been
       destroyed.  This is an inherent limitation of frame pointers.

b) ORC (Oops Rewind Capability) unwind table generation

   An alternative to frame pointers and DWARF, ORC unwind data can be
   used to walk the stack.  Unlike frame pointers, ORC data is out of
   band.  So it doesn't affect runtime performance and it can be
   reliable even when interrupts or exceptions are involved.

   For more details, see Documentation/x86/orc-unwinder.rst.

c) Higher live patching compatibility rate

   Livepatch has an optional "consistency model", which is needed for
   more complex patches.  In order for the consistency model to work,
   stack traces need to be reliable (or an unreliable condition needs to
   be detectable).  Objtool makes that possible.

   For more details, see the livepatch documentation in the Linux kernel
   source tree at Documentation/livepatch/livepatch.rst.

Rules
-----

To achieve the validation, objtool enforces the following rules:

1. Each callable function must be annotated as such with the ELF
   function type.  In asm code, this is typically done using the
   ENTRY/ENDPROC macros.  If objtool finds a return instruction
   outside of a function, it flags an error since that usually indicates
   callable code which should be annotated accordingly.

   This rule is needed so that objtool can properly identify each
   callable function in order to analyze its stack metadata.

2. Conversely, each section of code which is *not* callable should *not*
   be annotated as an ELF function.  The ENDPROC macro shouldn't be used
   in this case.

   This rule is needed so that objtool can ignore non-callable code.
   Such code doesn't have to follow any of the other rules.

3. Each callable function which calls another function must have the
   correct frame pointer logic, if required by CONFIG_FRAME_POINTER or
   the architecture's back chain rules.  This can by done in asm code
   with the FRAME_BEGIN/FRAME_END macros.

   This rule ensures that frame pointer based stack traces will work as
   designed.  If function A doesn't create a stack frame before calling
   function B, the _caller_ of function A will be skipped on the stack
   trace.

4. Dynamic jumps and jumps to undefined symbols are only allowed if:

   a) the jump is part of a switch statement; or

   b) the jump matches sibling call semantics and the frame pointer has
      the same value it had on function entry.

   This rule is needed so that objtool can reliably analyze all of a
   function's code paths.  If a function jumps to code in another file,
   and it's not a sibling call, objtool has no way to follow the jump
   because it only analyzes a single file at a time.

5. A callable function may not execute kernel entry/exit instructions.
   The only code which needs such instructions is kernel entry code,
   which shouldn't be be in callable functions anyway.

   This rule is just a sanity check to ensure that callable functions
   return normally.


Objtool warnings
----------------

For asm files, if you're getting an error which doesn't make sense,
first make sure that the affected code follows the above rules.

For C files, the common culprits are inline asm statements and calls to
"noreturn" functions.  See below for more details.

Another possible cause for errors in C code is if the Makefile removes
-fno-omit-frame-pointer or adds -fomit-frame-pointer to the gcc options.

Here are some examples of common warnings reported by objtool, what
they mean, and suggestions for how to fix them.


1. file.o: warning: objtool: func()+0x128: call without frame pointer save/setup

   The func() function made a function call without first saving and/or
   updating the frame pointer, and CONFIG_FRAME_POINTER is enabled.

   If the error is for an asm file, and func() is indeed a callable
   function, add proper frame pointer logic using the FRAME_BEGIN and
   FRAME_END macros.  Otherwise, if it's not a callable function, remove
   its ELF function annotation by changing ENDPROC to END, and instead
   use the manual unwind hint macros in asm/unwind_hints.h.

   If it's a GCC-compiled .c file, the error may be because the function
   uses an inline asm() statement which has a "call" instruction.  An
   asm() statement with a call instruction must declare the use of the
   stack pointer in its output operand.  On x86_64, this means adding
   the ASM_CALL_CONSTRAINT as an output constraint:

     asm volatile("call func" : ASM_CALL_CONSTRAINT);

   Otherwise the stack frame may not get created before the call.


2. file.o: warning: objtool: .text+0x53: unreachable instruction

   Objtool couldn't find a code path to reach the instruction.

   If the error is for an asm file, and the instruction is inside (or
   reachable from) a callable function, the function should be annotated
   with the ENTRY/ENDPROC macros (ENDPROC is the important one).
   Otherwise, the code should probably be annotated with the unwind hint
   macros in asm/unwind_hints.h so objtool and the unwinder can know the
   stack state associated with the code.

   If you're 100% sure the code won't affect stack traces, or if you're
   a just a bad person, you can tell objtool to ignore it.  See the
   "Adding exceptions" section below.

   If it's not actually in a callable function (e.g. kernel entry code),
   change ENDPROC to END.


4. file.o: warning: objtool: func(): can't find starting instruction
   or
   file.o: warning: objtool: func()+0x11dd: can't decode instruction

   Does the file have data in a text section?  If so, that can confuse
   objtool's instruction decoder.  Move the data to a more appropriate
   section like .data or .rodata.


5. file.o: warning: objtool: func()+0x6: unsupported instruction in callable function

   This is a kernel entry/exit instruction like sysenter or iret.  Such
   instructions aren't allowed in a callable function, and are most
   likely part of the kernel entry code.  They should usually not have
   the callable function annotation (ENDPROC) and should always be
   annotated with the unwind hint macros in asm/unwind_hints.h.


6. file.o: warning: objtool: func()+0x26: sibling call from callable instruction with modified stack frame

   This is a dynamic jump or a jump to an undefined symbol.  Objtool
   assumed it's a sibling call and detected that the frame pointer
   wasn't first restored to its original state.

   If it's not really a sibling call, you may need to move the
   destination code to the local file.

   If the instruction is not actually in a callable function (e.g.
   kernel entry code), change ENDPROC to END and annotate manually with
   the unwind hint macros in asm/unwind_hints.h.


7. file: warning: objtool: func()+0x5c: stack state mismatch

   The instruction's frame pointer state is inconsistent, depending on
   which execution path was taken to reach the instruction.

   Make sure that, when CONFIG_FRAME_POINTER is enabled, the function
   pushes and sets up the frame pointer (for x86_64, this means rbp) at
   the beginning of the function and pops it at the end of the function.
   Also make sure that no other code in the function touches the frame
   pointer.

   Another possibility is that the code has some asm or inline asm which
   does some unusual things to the stack or the frame pointer.  In such
   cases it's probably appropriate to use the unwind hint macros in
   asm/unwind_hints.h.


8. file.o: warning: objtool: funcA() falls through to next function funcB()

   This means that funcA() doesn't end with a return instruction or an
   unconditional jump, and that objtool has determined that the function
   can fall through into the next function.  There could be different
   reasons for this:

   1) funcA()'s last instruction is a call to a "noreturn" function like
      panic().  In this case the noreturn function needs to be added to
      objtool's hard-coded global_noreturns array.  Feel free to bug the
      objtool maintainer, or you can submit a patch.

   2) funcA() uses the unreachable() annotation in a section of code
      that is actually reachable.

   3) If funcA() calls an inline function, the object code for funcA()
      might be corrupt due to a gcc bug.  For more details, see:
      https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70646

9. file.o: warning: objtool: funcA() call to funcB() with UACCESS enabled

   This means that an unexpected call to a non-whitelisted function exists
   outside of arch-specific guards.
   X86: SMAP (stac/clac): __uaccess_begin()/__uaccess_end()
   ARM: PAN: uaccess_enable()/uaccess_disable()

   These functions should be called to denote a minimal critical section around
   access to __user variables. See also: https://lwn.net/Articles/517475/

   The intention of the warning is to prevent calls to funcB() from eventually
   calling schedule(), potentially leaking the AC flags state, and not
   restoring them correctly.

   It also helps verify that there are no unexpected calls to funcB() which may
   access user space pages with protections against doing so disabled.

   To fix, either:
   1) remove explicit calls to funcB() from funcA().
   2) add the correct guards before and after calls to low level functions like
      __get_user_size()/__put_user_size().
   3) add funcB to uaccess_safe_builtin whitelist in tools/objtool/check.c, if
      funcB obviously does not call schedule(), and is marked notrace (since
      function tracing inserts additional calls, which is not obvious from the
      sources).

10. file.o: warning: func()+0x5c: stack layout conflict in alternatives

    This means that in the use of the alternative() or ALTERNATIVE()
    macro, the code paths have conflicting modifications to the stack.
    The problem is that there is only one ORC unwind table, which means
    that the ORC unwind entries must be consistent for all possible
    instruction boundaries regardless of which code has been patched.
    This limitation can be overcome by massaging the alternatives with
    NOPs to shift the stack changes around so they no longer conflict.

11. file.o: warning: unannotated intra-function call

   This warning means that a direct call is done to a destination which
   is not at the beginning of a function. If this is a legit call, you
   can remove this warning by putting the ANNOTATE_INTRA_FUNCTION_CALL
   directive right before the call.


If the error doesn't seem to make sense, it could be a bug in objtool.
Feel free to ask the objtool maintainer for help.


Adding exceptions
-----------------

If you _really_ need objtool to ignore something, and are 100% sure
that it won't affect kernel stack traces, you can tell objtool to
ignore it:

- To skip validation of a function, use the STACK_FRAME_NON_STANDARD
  macro.

- To skip validation of a file, add

    OBJECT_FILES_NON_STANDARD_filename.o := y

  to the Makefile.

- To skip validation of a directory, add

    OBJECT_FILES_NON_STANDARD := y

  to the Makefile.