1CRASH(8) System Manager's Manual CRASH(8)
2
6 crash - Analyze Linux crash dump data or a live system
7
9 crash [OPTION]... NAMELIST MEMORY-IMAGE[@ADDRESS] (dumpfile form)
10 crash [OPTION]... [NAMELIST] (live system form)
11
13 Crash is a tool for interactively analyzing the state of the Linux sys‐
14 tem while it is running, or after a kernel crash has occurred and a
15 core dump has been created by the netdump, diskdump, LKCD, kdump, xen‐
16 dump or kvmdump facilities. It is loosely based on the SVR4 UNIX crash
17 command, but has been significantly enhanced by completely merging it
18 with the gdb(1) debugger. The marriage of the two effectively combines
19 the kernel-specific nature of the traditional UNIX crash utility with
20 the source code level debugging capabilities of gdb(1).
21 In the dumpfile form, both a NAMELIST and a MEMORY-IMAGE argument must
23 be entered. In the live system form, the NAMELIST argument must be
24 entered if the kernel's vmlinux file is not located in a known loca‐
25 tion, such as the /usr/lib/debug/lib/modules/<kernel-version> direc‐
26 tory.
27 The crash utility has also been extended to support the analysis of
29 dumpfiles generated by a crash of the Xen hypervisor. In that case,
30 the NAMELIST argument must be that of the xen-syms binary. Live system
31 analysis is not supported for the Xen hypervisor.
32 The crash utility command set consists of common kernel core analysis
34 tools such as kernel stack back traces of all processes, source code
35 disassembly, formatted kernel structure and variable displays, virtual
36 memory data, dumps of linked-lists, etc., along with several commands
37 that delve deeper into specific kernel subsystems. Appropriate gdb
38 commands may also be entered, which in turn are passed on to the gdb
39 module for execution. If desired, commands may be placed in either a
40 $HOME/.crashrc file and/or in a .crashrc file in the current directory.
41 During initialization, the commands in $HOME/.crashrc are executed
42 first, followed by those in the ./.crashrc file.
43 The crash utility is designed to be independent of Linux version depen‐
45 dencies. When new kernel source code impacts the correct functionality
46 of crash and its command set, the utility will be updated to recognize
47 new kernel code changes, while maintaining backwards compatibility with
48 earlier releases.
49
51 NAMELIST
52 This is a pathname to an uncompressed kernel image (a vmlinux
53 file), or a Xen hypervisor image (a xen-syms file) which has
54 been compiled with the "-g" option. If using the dumpfile form,
55 a vmlinux file may be compressed in either gzip or bzip2 for‐
56 mats.
57 MEMORY-IMAGE[@ADDRESS]
59 A kernel core dump file created by the netdump, diskdump, LKCD
60 kdump, xendump or kvmdump facilities.
61 If a MEMORY-IMAGE argument is not entered, the session will be
63 invoked on the live system, which typically requires root privi‐
64 leges because of the device file used to access system RAM. By
65 default, /dev/crash will be used if it exists. If it does not
66 exist, then /dev/mem will be used; but if the kernel has been
67 configured with CONFIG_STRICT_DEVMEM, then /proc/kcore will be
68 used. It is permissible to explicitly enter /dev/crash,
69 /dev/mem or /proc/kcore.
70 An @ADDRESS value must be appended to the MEMORY-IMAGE if the
72 dumpfile is a raw RAM dumpfile that has no header information
73 describing the file contents. Multiple MEMORY-IMAGE@ADDRESS
74 ordered pairs may be entered, with each dumpfile containing a
75 contiguous block of RAM, where the ADDRESS value is the physical
76 start address of the block expressed in hexadecimal. The physi‐
77 cal address value(s) will be used to create a temporary ELF
78 header in /var/tmp, which will only exist during the crash ses‐
79 sion. If a raw RAM dumpile represents a live memory source,
80 such as that specified by the QEMU mem-path argument of a mem‐
81 ory-backend-file object, then "live:" must be prepended to the
82 MEMORY-IMAGE name.
83 mapfile
85 If the NAMELIST file is not the same kernel that is running
86 (live system form), or the kernel that was running when the sys‐
87 tem crashed (dumpfile form), then the System.map file of the
88 original kernel should be entered on the command line.
89 -h [option]
91 --help [option]
92 Without an option argument, display a crash usage help message.
93 If the option argument is a crash command name, the help page
94 for that command is displayed. If it is the string "input", a
95 page describing the various crash command line input options is
96 displayed. If it is the string "output", a page describing com‐
97 mand line output options is displayed. If it is the string
98 "all", then all of the possible help messages are displayed.
99 After the help message is displayed, crash exits.
100 -s Silently proceed directly to the "crash>" prompt without dis‐
102 playing any version, GPL, or crash initialization data during
103 startup, and by default, runtime command output is not passed to
104 any scrolling command.
105 -i file
107 Execute the command(s) contained in file prior to displaying the
108 "crash>" prompt for interactive user input.
109 -d num Set the internal debug level. The higher the number, the more
111 debugging data will be printed when crash initializes and runs.
112 -S Use /boot/System.map as the mapfile.
114 -e vi | emacs
116 Set the readline(3) command line editing mode to "vi" or
117 "emacs". The default editing mode is "vi".
118 -f Force the usage of a compressed vmlinux file if its original
120 name does not start with "vmlinux".
121 -k Indicate that the NAMELIST file is an LKCD "Kerntypes" debuginfo
123 file.
124 -g [namelist]
126 Determine if a vmlinux or xen-syms namelist file contains debug‐
127 ging data.
128 -t Display the system-crash timestamp and exit.
130 -L Attempt to lock all of its virtual address space into memory by
132 calling mlockall(MCL_CURRENT|MCL_FUTURE) during initialization.
133 If the system call fails, an error message will be displayed,
134 but the session continues.
135 -c tty-device
137 Open the tty-device as the console used for debug messages.
138 -p page-size
140 If a processor's page size cannot be determined by the dumpfile,
141 and the processor default cannot be used, use page-size.
142 -o filename
144 Only used with the MEMORY-IMAGE@ADDRESS format for raw RAM dump‐
145 files, specifies a filename of a new ELF vmcore that will be
146 created and used as the dumpfile. It will be saved to allow
147 future use as a standalone vmcore, replacing the original raw
148 RAM dumpfile.
149 -m option=value
151 --machdep option=value
152 Pass an option and value pair to machine-dependent code. These
153 architecture-specific option/pairs should only be required in
154 very rare circumstances:
155 X86_64:
157 phys_base=<physical-address>
158 irq_eframe_link=<value>
159 irq_stack_gap=<value>
160 max_physmem_bits=<value>
161 kernel_image_size=<value>
162 vm=orig (pre-2.6.11 virtual memory address ranges)
163 vm=2.6.11 (2.6.11 and later virtual memory address ranges)
164 vm=xen (Xen kernel virtual memory address ranges)
165 vm=xen-rhel4 (RHEL4 Xen kernel virtual address ranges)
166 vm=5level (5-level page tables)
167 PPC64:
168 vm=orig
169 vm=2.6.14 (4-level page tables)
170 IA64:
171 phys_start=<physical-address>
172 init_stack_size=<size>
173 vm=4l (4-level page tables)
174 ARM:
175 phys_base=<physical-address>
176 ARM64:
177 phys_offset=<physical-address>
178 kimage_voffset=<kimage_voffset-value>
179 max_physmem_bits=<value>
180 X86:
181 page_offset=<CONFIG_PAGE_OFFSET-value>
182 -x Automatically load extension modules from a particular direc‐
184 tory. If a directory is specified in the CRASH_EXTENSIONS shell
185 environment variable, then that directory will be used. Other‐
186 wise /usr/lib64/crash/extensions (64-bit architectures) or
187 /usr/lib/crash/extensions (32-bit architectures) will be used;
188 if they do not exist, then the ./extensions directory will be
189 used.
190 --active
192 Track only the active task on each cpu.
193 --buildinfo
195 Display the crash binary's build date, the user ID of the
196 builder, the hostname of the machine where the build was done,
197 the target architecture, the version number, and the compiler
198 version.
199 --memory_module modname
201 Use the modname as an alternative kernel module to the crash.ko
202 module that creates the /dev/crash device.
203 --memory_device device
205 Use device as an alternative device to the /dev/crash, /dev/mem
206 or /proc/kcore devices.
207 --log dumpfile
209 Dump the contents of the kernel log buffer. A kernel namelist
210 argument is not necessary, but the dumpfile must contain the
211 VMCOREINFO data taken from the original /proc/vmcore ELF header.
212 --no_kallsyms
214 Do not use kallsyms-generated symbol information contained
215 within kernel module object files.
216 --no_modules
218 Do not access or display any kernel module related information.
219 --no_ikconf
221 Do not attempt to read configuration data that was built into
222 kernels configured with CONFIG_IKCONFIG.
223 --no_data_debug
225 Do not verify the validity of all structure member offsets and
226 structure sizes that it uses.
227 --no_kmem_cache
229 Do not initialize the kernel's slab cache infrastructure, and
230 commands that use kmem_cache-related data will not work.
231 --no_elf_notes
233 Do not use the registers from the ELF NT_PRSTATUS notes saved in
234 a compressed kdump header for backtraces.
235 --kmem_cache_delay
237 Delay the initialization of the kernel's slab cache infrastruc‐
238 ture until it is required by a run-time command.
239 --readnow
241 Pass this flag to the embedded gdb module, which will override
242 its two-stage strategy that it uses for reading symbol tables
243 from the NAMELIST.
244 --smp Specify that the system being analyzed is an SMP kernel.
246 -v
248 --version
249 Display the version of the crash utility, the version of the
250 embedded gdb module, GPL information, and copyright notices.
251 --cpus number
253 Specify the number of cpus in the SMP system being analyzed.
254 --osrelease dumpfile
256 Display the OSRELEASE vmcoreinfo string from a kdump dumpfile
257 header.
258 --hyper
260 Force the session to be that of a Xen hypervisor.
261 --p2m_mfn pfn
263 When a Xen Hypervisor or its dom0 kernel crashes, the dumpfile
264 is typically analyzed with either the Xen hypervisor or the dom0
265 kernel. It is also possible to analyze any of the guest domU
266 kernels if the pfn_to_mfn_list_list pfn value of the guest ker‐
267 nel is passed on the command line along with its NAMELIST and
268 the dumpfile.
269 --xen_phys_start physical-address
271 Supply the base physical address of the Xen hypervisor's text
272 and static data for older xendump dumpfiles that did not pass
273 that information in the dumpfile header.
274 --zero_excluded
276 If the makedumpfile(8) facility has filtered a compressed kdump
277 dumpfile to exclude various types of non-essential pages, or has
278 marked a compressed or ELF kdump dumpfile as incomplete due to
279 an ENOSPC or other error during its creation, any attempt to
280 read missing pages will fail. With this flag, reads from any of
281 those pages will return zero-filled memory.
282 --no_panic
284 Do not attempt to find the task that was running when the kernel
285 crashed. Set the initial context to that of the "swapper" task
286 on cpu 0.
287 --more Use /bin/more as the command output scroller, overriding the
289 default of /usr/bin/less and any settings in either ./.crashrc
290 or $HOME/.crashrc.
291 --less Use /usr/bin/less as the command output scroller, overriding any
293 settings in either ./.crashrc or $HOME/.crashrc.
294 --hex Set the default command output radix to 16, overriding the
296 default radix of 10, and any radix settings in either ./.crashrc
297 or $HOME/.crashrc.
298 --dec Set the default command output radix to 10, overriding any radix
300 settings in either ./.crashrc or $HOME/.crashrc. This is the
301 default radix setting.
302 --CRASHPAGER
304 Use the output paging command defined in the CRASHPAGER shell
305 environment variable, overriding any settings in either
306 ./.crashrc or $HOME/.crashrc.
307 --no_scroll
309 Do not pass run-time command output to any scrolling command.
310 --no_strip
312 Do not strip cloned kernel text symbol names.
313 --no_crashrc
315 Do not execute the commands in either $HOME/.crashrc or
316 ./.crashrc.
317 --mod directory
319 When loading the debuginfo data of kernel modules with the mod
320 -S command, search for their object files in directory instead
321 of in the standard location.
322 --kaslr offset|auto
324 If an x86_64 kernel was configured with CONFIG_RANDOMIZE_BASE,
325 the offset value is equal to the difference between the symbol
326 values compiled into the vmlinux file and their relocated KASLR
327 values. If set to auto, the KASLR offset value will be automat‐
328 ically calculated.
329 --reloc size
331 When analyzing live x86 kernels that were configured with a CON‐
332 FIG_PHYSICAL_START value that is larger than its CONFIG_PHYSI‐
333 CAL_ALIGN value, then it will be necessary to enter a relocation
334 size equal to the difference between the two values.
335 --hash count
337 Set the number of internal hash queue heads used for list gath‐
338 ering and verification. The default count is 32768.
339 --minimal
341 Bring up a session that is restricted to the log, dis, rd, sym,
342 eval, set and exit commands. This option may provide a way to
343 extract some minimal/quick information from a corrupted or trun‐
344 cated dumpfile, or in situations where one of the several kernel
345 subsystem initialization routines would abort the crash session.
346 --kvmhost [32|64]
348 When examining an x86 KVM guest dumpfile, this option specifies
349 that the KVM host that created the dumpfile was an x86 (32-bit)
350 or an x86_64 (64-bit) machine, overriding the automatically
351 determined value.
352 --kvmio <size>
354 override the automatically-calculated KVM guest I/O hole size.
355 --offline [show|hide]
357 Show or hide command output that is related to offline cpus.
358 The default setting is show.
359
361 Each crash command generally falls into one of the following cate‐
362 gories:
363 Symbolic display
365 Displays of kernel text/data, which take full advantage of the
366 power of gdb to format and display data structures symbolically.
367 System state
369 The majority of crash commands consist of a set of "kernel-
370 aware" commands, which delve into various kernel subsystems on a
371 system-wide or per-task basis.
372 Utility functions
374 A set of useful helper commands serving various purposes, some
375 simple, others quite powerful.
376 Session control
378 Commands that control the crash session itself.
379 The following alphabetical list consists of a very simple overview of
381 each crash command. However, since individual commands often have sev‐
382 eral options resulting in significantly different output, it is sug‐
383 gested that the full description of each command be viewed by executing
384 crash -h <command>, or during a crash session by simply entering help
385 command.
386 * "pointer to" is shorthand for either the struct or union com‐
388 mands. It displays the contents of a kernel structure or union.
389 alias creates a single-word alias for a command.
391 ascii displays an ascii chart or translates a numeric value into its
393 ascii components.
394 bt displays a task's kernel-stack backtrace. If it is given the -a
396 option, it displays the stack traces of the active tasks on all
397 CPUs. It is often used with the foreach command to display the
398 backtraces of all tasks with one command.
399 btop translates a byte value (physical offset) to its page number.
401 dev displays data concerning the character and block device assign‐
403 ments, I/O port usage, I/O memory usage, and PCI device data.
404 dis disassembles memory, either entire kernel functions, from a
406 location for a specified number of instructions, or from the
407 start of a function up to a specified memory location.
408 eval evaluates an expression or numeric type and displays the result
410 in hexadecimal, decimal, octal and binary.
411 exit causes crash to exit.
413 extend dynamically loads or unloads crash shared object extension mod‐
415 ules.
416 files displays information about open files in a context.
418 foreach
420 repeats a specified command for the specified (or all) tasks in
421 the system.
422 fuser displays the tasks using the specified file or socket.
424 gdb passes its argument to the embedded gdb module. It is useful
426 for executing gdb commands that have the same name as crash com‐
427 mands.
428 help alone displays the command menu; if followed by a command name,
430 a full description of a command, its options, and examples are
431 displayed. Its output is far more complete and useful than this
432 man page.
433 ipcs displays data about the System V IPC facilities.
435 irq displays data concerning interrupt request numbers and bottom-
437 half interrupt handling.
438 kmem displays information about the use of kernel memory.
440 list displays the contents of a linked list.
442 log displays the kernel log_buf contents in chronological order.
444 mach displays data specific to the machine type.
446 mod displays information about the currently installed kernel mod‐
448 ules, or adds or deletes symbolic or debugging information about
449 specified kernel modules.
450 mount displays information about the currently-mounted filesystems.
452 net display various network related data.
454 p passes its arguments to the gdb "print" command for evaluation
456 and display.
457 ps displays process status for specified, or all, processes in the
459 system.
460 pte translates the hexadecimal contents of a PTE into its physical
462 page address and page bit settings.
463 ptob translates a page frame number to its byte value.
465 ptov translates a hexadecimal physical address into a kernel virtual
467 address.
468 q is an alias for the "exit" command.
470 rd displays the contents of memory, with the output formatted in
472 several different manners.
473 repeat repeats a command indefinitely, optionally delaying a given num‐
475 ber of seconds between each command execution.
476 runq displays the tasks on the run queue.
478 search searches a range of user or kernel memory space for given value.
480 set either sets a new context, or gets the current context for dis‐
482 play.
483 sig displays signal-handling data of one or more tasks.
485 struct displays either a structure definition or the contents of a ker‐
487 nel structure at a specified address.
488 swap displays information about each configured swap device.
490 sym translates a symbol to its virtual address, or a static kernel
492 virtual address to its symbol -- or to a symbol-plus-offset
493 value, if appropriate.
494 sys displays system-specific data.
496 task displays the contents of a task_struct.
498 tree displays the contents of a red-black tree or a radix tree.
500 timer displays the timer queue entries, both old- and new-style, in
502 chronological order.
503 union is similar to the struct command, except that it works on kernel
505 unions.
506 vm displays basic virtual memory information of a context.
508 vtop translates a user or kernel virtual address to its physical
510 address.
511 waitq walks the wait queue list displaying the tasks which are blocked
513 on the specified wait queue.
514 whatis displays the definition of structures, unions, typedefs or
516 text/data symbols.
517 wr modifies the contents of memory on a live system. It can only
519 be used if /dev/mem is the device file being used to access sys‐
520 tem RAM, and should obviously be used with great care.
521 When crash is invoked with a Xen hypervisor binary as the NAMELIST, the
523 command set is slightly modified. The *, alias, ascii, bt, dis, eval,
524 exit, extend, gdb, help, list, log, p, pte, rd, repeat, search, set,
525 struct, sym, sys, union, whatis, wr and q commands are the same as
526 above. The following commands are specific to the Xen hypervisor:
527 domain displays the contents of the domain structure for selected, or
529 all, domains.
530 doms displays domain status for selected, or all, domains.
532 dumpinfo
534 displays Xen dump information for selected, or all, cpus.
535 pcpus displays physical cpu information for selected, or all, cpus.
537 vcpus displays vcpu status for selected, or all, vcpus.
539
541 .crashrc
542 Initialization commands. The file can be located in the user's
543 HOME directory and/or the current directory. Commands found in
544 the .crashrc file in the HOME directory are executed before
545 those in the current directory's .crashrc file.
546
548 EDITOR Command input is read using readline(3). If EDITOR is set to
549 emacs or vi then suitable keybindings are used. If EDITOR is
550 not set, then vi is used. This can be overridden by set vi or
551 set emacs commands located in a .crashrc file, or by entering -e
552 emacs on the crash command line.
553 CRASHPAGER
555 If CRASHPAGER is set, its value is used as the name of the pro‐
556 gram to which command output will be sent. If not, then command
557 output is sent to /usr/bin/less -E -X by default.
558 CRASH_MODULE_PATH
560 Specifies an alternative directory tree to search for kernel
561 module object files.
562 CRASH_EXTENSIONS
564 Specifies a directory containing extension modules that will be
565 loaded automatically if the -x command line option is used.
566
568 If crash does not work, look for a newer version: kernel evolution fre‐
569 quently makes crash updates necessary.
570 The command set scroll off will cause output to be sent directly to the
572 terminal rather than through a paging program. This is useful, for
573 example, if you are running crash in a window of emacs.
574
576 Dave Anderson <anderson@redhat.com> wrote crash.
577 Jay Fenlason <fenlason@redhat.com> and Dave Anderson <anderson@red‐
579 hat.com> wrote this man page.
580
582 The help command within crash provides more complete and accurate docu‐
583 mentation than this man page.
584 http://people.redhat.com/anderson - the home page of the crash utility.
586 netdump(8), gdb(1), makedumpfile(8)
588 CRASH(8)