6 ``ksymoops`` is useless on 2.6 or upper. Please use the Oops in its original
7 format (from ``dmesg``, etc). Ignore any references in this or other docs to
8 "decoding the Oops" or "running it through ksymoops".
9 If you post an Oops from 2.6+ that has been run through ``ksymoops``,
10 people will just tell you to repost it.
15 Find the Oops and send it to the maintainer of the kernel area that seems to be
16 involved with the problem. Don't worry too much about getting the wrong person.
17 If you are unsure send it to the person responsible for the code relevant to
18 what you were doing. If it occurs repeatably try and describe how to recreate
19 it. That's worth even more than the oops.
21 If you are totally stumped as to whom to send the report, send it to
22 linux-kernel@vger.kernel.org. Thanks for your help in making Linux as
23 stable as humanly possible.
26 ----------------------
28 Normally the Oops text is read from the kernel buffers by klogd and
29 handed to ``syslogd`` which writes it to a syslog file, typically
30 ``/var/log/messages`` (depends on ``/etc/syslog.conf``). Sometimes ``klogd``
31 dies, in which case you can run ``dmesg > file`` to read the data from the
32 kernel buffers and save it. Or you can ``cat /proc/kmsg > file``, however you
33 have to break in to stop the transfer, ``kmsg`` is a "never ending file".
34 If the machine has crashed so badly that you cannot enter commands or
35 the disk is not available then you have three options :
37 (1) Hand copy the text from the screen and type it in after the machine
38 has restarted. Messy but it is the only option if you have not
39 planned for a crash. Alternatively, you can take a picture of
40 the screen with a digital camera - not nice, but better than
41 nothing. If the messages scroll off the top of the console, you
42 may find that booting with a higher resolution (eg, ``vga=791``)
43 will allow you to read more of the text. (Caveat: This needs ``vesafb``,
44 so won't help for 'early' oopses)
46 (2) Boot with a serial console (see
47 :ref:`Documentation/serial-console.txt <serial_console>`),
48 run a null modem to a second machine and capture the output there
49 using your favourite communication program. Minicom works well.
51 (3) Use Kdump (see Documentation/kdump/kdump.txt),
52 extract the kernel ring buffer from old memory with using dmesg
53 gdbmacro in Documentation/kdump/gdbmacros.txt.
61 the message from Linus below applies to 2.4 kernel. I have preserved it
62 for historical reasons, and because some of the information in it still
63 applies. Especially, please ignore any references to ksymoops.
67 From: Linus Torvalds <torvalds@osdl.org>
69 How to track down an Oops.. [originally a mail to linux-kernel]
71 The main trick is having 5 years of experience with those pesky oops
74 Actually, there are things you can do that make this easier. I have two
77 gdb /usr/src/linux/vmlinux
78 gdb> disassemble <offending_function>
80 That's the easy way to find the problem, at least if the bug-report is
81 well made (like this one was - run through ``ksymoops`` to get the
82 information of which function and the offset in the function that it
85 Oh, it helps if the report happens on a kernel that is compiled with the
86 same compiler and similar setups.
88 The other thing to do is disassemble the "Code:" part of the bug report:
89 ksymoops will do this too with the correct tools, but if you don't have
90 the tools you can just do a silly program::
92 char str[] = "\xXX\xXX\xXX...";
95 and compile it with ``gcc -g`` and then do ``disassemble str`` (where the ``XX``
96 stuff are the values reported by the Oops - you can just cut-and-paste
97 and do a replace of spaces to ``\x`` - that's what I do, as I'm too lazy
98 to write a program to automate this all).
100 Alternatively, you can use the shell script in ``scripts/decodecode``.
103 decodecode < oops.txt
105 The hex bytes that follow "Code:" may (in some architectures) have a series
106 of bytes that precede the current instruction pointer as well as bytes at and
107 following the current instruction pointer. In some cases, one instruction
108 byte or word is surrounded by ``<>`` or ``()``, as in ``<86>`` or ``(f00d)``.
109 These ``<>`` or ``()`` markings indicate the current instruction pointer.
111 Example from i386, split into multiple lines for readability::
113 Code: f9 0f 8d f9 00 00 00 8d 42 0c e8 dd 26 11 c7 a1 60 ea 2b f9 8b 50 08 a1
114 64 ea 2b f9 8d 34 82 8b 1e 85 db 74 6d 8b 15 60 ea 2b f9 <8b> 43 04 39 42 54
115 7e 04 40 89 42 54 8b 43 04 3b 05 00 f6 52 c0
117 Finally, if you want to see where the code comes from, you can do::
120 make fs/buffer.s # or whatever file the bug happened in
122 and then you get a better idea of what happens than with the gdb
125 Now, the trick is just then to combine all the data you have: the C
126 sources (and general knowledge of what it **should** do), the assembly
127 listing and the code disassembly (and additionally the register dump you
128 also get from the "oops" message - that can be useful to see **what** the
129 corrupted pointers were, and when you have the assembler listing you can
130 also match the other registers to whatever C expressions they were used
133 Essentially, you just look at what doesn't match (in this case it was the
134 "Code" disassembly that didn't match with what the compiler generated).
135 Then you need to find out **why** they don't match. Often it's simple - you
136 see that the code uses a NULL pointer and then you look at the code and
137 wonder how the NULL pointer got there, and if it's a valid thing to do
138 you just check against it..
140 Now, if somebody gets the idea that this is time-consuming and requires
141 some small amount of concentration, you're right. Which is why I will
142 mostly just ignore any panic reports that don't have the symbol table
143 info etc looked up: it simply gets too hard to look it up (I have some
144 programs to search for specific patterns in the kernel code segment, and
145 sometimes I have been able to look up those kinds of panics too, but
146 that really requires pretty good knowledge of the kernel just to be able
147 to pick out the right sequences etc..)
149 **Sometimes** it happens that I just see the disassembled code sequence
150 from the panic, and I know immediately where it's coming from. That's when
151 I get worried that I've been doing this for too long ;-)
156 ---------------------------------------------------------------------------
158 Notes on Oops tracing with ``klogd``
159 ------------------------------------
161 In order to help Linus and the other kernel developers there has been
162 substantial support incorporated into ``klogd`` for processing protection
163 faults. In order to have full support for address resolution at least
164 version 1.3-pl3 of the ``sysklogd`` package should be used.
166 When a protection fault occurs the ``klogd`` daemon automatically
167 translates important addresses in the kernel log messages to their
168 symbolic equivalents. This translated kernel message is then
169 forwarded through whatever reporting mechanism ``klogd`` is using. The
170 protection fault message can be simply cut out of the message files
171 and forwarded to the kernel developers.
173 Two types of address resolution are performed by ``klogd``. The first is
174 static translation and the second is dynamic translation. Static
175 translation uses the System.map file in much the same manner that
176 ksymoops does. In order to do static translation the ``klogd`` daemon
177 must be able to find a system map file at daemon initialization time.
178 See the klogd man page for information on how ``klogd`` searches for map
181 Dynamic address translation is important when kernel loadable modules
182 are being used. Since memory for kernel modules is allocated from the
183 kernel's dynamic memory pools there are no fixed locations for either
184 the start of the module or for functions and symbols in the module.
186 The kernel supports system calls which allow a program to determine
187 which modules are loaded and their location in memory. Using these
188 system calls the klogd daemon builds a symbol table which can be used
189 to debug a protection fault which occurs in a loadable kernel module.
191 At the very minimum klogd will provide the name of the module which
192 generated the protection fault. There may be additional symbolic
193 information available if the developer of the loadable module chose to
194 export symbol information from the module.
196 Since the kernel module environment can be dynamic there must be a
197 mechanism for notifying the ``klogd`` daemon when a change in module
198 environment occurs. There are command line options available which
199 allow klogd to signal the currently executing daemon that symbol
200 information should be refreshed. See the ``klogd`` manual page for more
203 A patch is included with the sysklogd distribution which modifies the
204 ``modules-2.0.0`` package to automatically signal klogd whenever a module
205 is loaded or unloaded. Applying this patch provides essentially
206 seamless support for debugging protection faults which occur with
207 kernel loadable modules.
209 The following is an example of a protection fault in a loadable module
210 processed by ``klogd``::
212 Aug 29 09:51:01 blizard kernel: Unable to handle kernel paging request at virtual address f15e97cc
213 Aug 29 09:51:01 blizard kernel: current->tss.cr3 = 0062d000, %cr3 = 0062d000
214 Aug 29 09:51:01 blizard kernel: *pde = 00000000
215 Aug 29 09:51:01 blizard kernel: Oops: 0002
216 Aug 29 09:51:01 blizard kernel: CPU: 0
217 Aug 29 09:51:01 blizard kernel: EIP: 0010:[oops:_oops+16/3868]
218 Aug 29 09:51:01 blizard kernel: EFLAGS: 00010212
219 Aug 29 09:51:01 blizard kernel: eax: 315e97cc ebx: 003a6f80 ecx: 001be77b edx: 00237c0c
220 Aug 29 09:51:01 blizard kernel: esi: 00000000 edi: bffffdb3 ebp: 00589f90 esp: 00589f8c
221 Aug 29 09:51:01 blizard kernel: ds: 0018 es: 0018 fs: 002b gs: 002b ss: 0018
222 Aug 29 09:51:01 blizard kernel: Process oops_test (pid: 3374, process nr: 21, stackpage=00589000)
223 Aug 29 09:51:01 blizard kernel: Stack: 315e97cc 00589f98 0100b0b4 bffffed4 0012e38e 00240c64 003a6f80 00000001
224 Aug 29 09:51:01 blizard kernel: 00000000 00237810 bfffff00 0010a7fa 00000003 00000001 00000000 bfffff00
225 Aug 29 09:51:01 blizard kernel: bffffdb3 bffffed4 ffffffda 0000002b 0007002b 0000002b 0000002b 00000036
226 Aug 29 09:51:01 blizard kernel: Call Trace: [oops:_oops_ioctl+48/80] [_sys_ioctl+254/272] [_system_call+82/128]
227 Aug 29 09:51:01 blizard kernel: Code: c7 00 05 00 00 00 eb 08 90 90 90 90 90 90 90 90 89 ec 5d c3
229 ---------------------------------------------------------------------------
233 Dr. G.W. Wettstein Oncology Research Div. Computing Facility
234 Roger Maris Cancer Center INTERNET: greg@wind.rmcc.com
240 ---------------------------------------------------------------------------
245 Some oops reports contain the string **'Tainted: '** after the program
246 counter. This indicates that the kernel has been tainted by some
247 mechanism. The string is followed by a series of position-sensitive
248 characters, each representing a particular tainted value.
250 1) 'G' if all modules loaded have a GPL or compatible license, 'P' if
251 any proprietary module has been loaded. Modules without a
252 MODULE_LICENSE or with a MODULE_LICENSE that is not recognised by
253 insmod as GPL compatible are assumed to be proprietary.
255 2) ``F`` if any module was force loaded by ``insmod -f``, ``' '`` if all
256 modules were loaded normally.
258 3) ``S`` if the oops occurred on an SMP kernel running on hardware that
259 hasn't been certified as safe to run multiprocessor.
260 Currently this occurs only on various Athlons that are not
263 4) ``R`` if a module was force unloaded by ``rmmod -f``, ``' '`` if all
264 modules were unloaded normally.
266 5) ``M`` if any processor has reported a Machine Check Exception,
267 ``' '`` if no Machine Check Exceptions have occurred.
269 6) ``B`` if a page-release function has found a bad page reference or
270 some unexpected page flags.
272 7) ``U`` if a user or user application specifically requested that the
273 Tainted flag be set, ``' '`` otherwise.
275 8) ``D`` if the kernel has died recently, i.e. there was an OOPS or BUG.
277 9) ``A`` if the ACPI table has been overridden.
279 10) ``W`` if a warning has previously been issued by the kernel.
280 (Though some warnings may set more specific taint flags.)
282 11) ``C`` if a staging driver has been loaded.
284 12) ``I`` if the kernel is working around a severe bug in the platform
285 firmware (BIOS or similar).
287 13) ``O`` if an externally-built ("out-of-tree") module has been loaded.
289 14) ``E`` if an unsigned module has been loaded in a kernel supporting
292 15) ``L`` if a soft lockup has previously occurred on the system.
294 16) ``K`` if the kernel has been live patched.
296 The primary reason for the **'Tainted: '** string is to tell kernel
297 debuggers if this is a clean kernel or if anything unusual has
298 occurred. Tainting is permanent: even if an offending module is
299 unloaded, the tainted value remains to indicate that the kernel is not