1================================================================
2Documentation for Kdump - The kexec-based Crash Dumping Solution
3================================================================
4
5This document includes overview, setup and installation, and analysis
6information.
7
8Overview
9========
10
11Kdump uses kexec to quickly boot to a dump-capture kernel whenever a
12dump of the system kernel's memory needs to be taken (for example, when
13the system panics). The system kernel's memory image is preserved across
14the reboot and is accessible to the dump-capture kernel.
15
16You can use common commands, such as cp and scp, to copy the
17memory image to a dump file on the local disk, or across the network to
18a remote system.
19
20Kdump and kexec are currently supported on the x86, x86_64, ppc64, ia64,
21s390x, arm and arm64 architectures.
22
23When the system kernel boots, it reserves a small section of memory for
24the dump-capture kernel. This ensures that ongoing Direct Memory Access
25(DMA) from the system kernel does not corrupt the dump-capture kernel.
26The kexec -p command loads the dump-capture kernel into this reserved
27memory.
28
29On x86 machines, the first 640 KB of physical memory is needed to boot,
30regardless of where the kernel loads. Therefore, kexec backs up this
31region just before rebooting into the dump-capture kernel.
32
33Similarly on PPC64 machines first 32KB of physical memory is needed for
34booting regardless of where the kernel is loaded and to support 64K page
35size kexec backs up the first 64KB memory.
36
37For s390x, when kdump is triggered, the crashkernel region is exchanged
38with the region [0, crashkernel region size] and then the kdump kernel
39runs in [0, crashkernel region size]. Therefore no relocatable kernel is
40needed for s390x.
41
42All of the necessary information about the system kernel's core image is
43encoded in the ELF format, and stored in a reserved area of memory
44before a crash. The physical address of the start of the ELF header is
45passed to the dump-capture kernel through the elfcorehdr= boot
46parameter. Optionally the size of the ELF header can also be passed
47when using the elfcorehdr=[size[KMG]@]offset[KMG] syntax.
48
49
50With the dump-capture kernel, you can access the memory image through
51/proc/vmcore. This exports the dump as an ELF-format file that you can
52write out using file copy commands such as cp or scp. Further, you can
53use analysis tools such as the GNU Debugger (GDB) and the Crash tool to
54debug the dump file. This method ensures that the dump pages are correctly
55ordered.
56
57
58Setup and Installation
59======================
60
61Install kexec-tools
62-------------------
63
641) Login as the root user.
65
662) Download the kexec-tools user-space package from the following URL:
67
68http://kernel.org/pub/linux/utils/kernel/kexec/kexec-tools.tar.gz
69
70This is a symlink to the latest version.
71
72The latest kexec-tools git tree is available at:
73
74- git://git.kernel.org/pub/scm/utils/kernel/kexec/kexec-tools.git
75- http://www.kernel.org/pub/scm/utils/kernel/kexec/kexec-tools.git
76
77There is also a gitweb interface available at
78http://www.kernel.org/git/?p=utils/kernel/kexec/kexec-tools.git
79
80More information about kexec-tools can be found at
81http://horms.net/projects/kexec/
82
833) Unpack the tarball with the tar command, as follows::
84
85	tar xvpzf kexec-tools.tar.gz
86
874) Change to the kexec-tools directory, as follows::
88
89	cd kexec-tools-VERSION
90
915) Configure the package, as follows::
92
93	./configure
94
956) Compile the package, as follows::
96
97	make
98
997) Install the package, as follows::
100
101	make install
102
103
104Build the system and dump-capture kernels
105-----------------------------------------
106There are two possible methods of using Kdump.
107
1081) Build a separate custom dump-capture kernel for capturing the
109   kernel core dump.
110
1112) Or use the system kernel binary itself as dump-capture kernel and there is
112   no need to build a separate dump-capture kernel. This is possible
113   only with the architectures which support a relocatable kernel. As
114   of today, i386, x86_64, ppc64, ia64, arm and arm64 architectures support
115   relocatable kernel.
116
117Building a relocatable kernel is advantageous from the point of view that
118one does not have to build a second kernel for capturing the dump. But
119at the same time one might want to build a custom dump capture kernel
120suitable to his needs.
121
122Following are the configuration setting required for system and
123dump-capture kernels for enabling kdump support.
124
125System kernel config options
126----------------------------
127
1281) Enable "kexec system call" in "Processor type and features."::
129
130	CONFIG_KEXEC=y
131
1322) Enable "sysfs file system support" in "Filesystem" -> "Pseudo
133   filesystems." This is usually enabled by default::
134
135	CONFIG_SYSFS=y
136
137   Note that "sysfs file system support" might not appear in the "Pseudo
138   filesystems" menu if "Configure standard kernel features (for small
139   systems)" is not enabled in "General Setup." In this case, check the
140   .config file itself to ensure that sysfs is turned on, as follows::
141
142	grep 'CONFIG_SYSFS' .config
143
1443) Enable "Compile the kernel with debug info" in "Kernel hacking."::
145
146	CONFIG_DEBUG_INFO=Y
147
148   This causes the kernel to be built with debug symbols. The dump
149   analysis tools require a vmlinux with debug symbols in order to read
150   and analyze a dump file.
151
152Dump-capture kernel config options (Arch Independent)
153-----------------------------------------------------
154
1551) Enable "kernel crash dumps" support under "Processor type and
156   features"::
157
158	CONFIG_CRASH_DUMP=y
159
1602) Enable "/proc/vmcore support" under "Filesystems" -> "Pseudo filesystems"::
161
162	CONFIG_PROC_VMCORE=y
163
164   (CONFIG_PROC_VMCORE is set by default when CONFIG_CRASH_DUMP is selected.)
165
166Dump-capture kernel config options (Arch Dependent, i386 and x86_64)
167--------------------------------------------------------------------
168
1691) On i386, enable high memory support under "Processor type and
170   features"::
171
172	CONFIG_HIGHMEM64G=y
173
174   or::
175
176	CONFIG_HIGHMEM4G
177
1782) On i386 and x86_64, disable symmetric multi-processing support
179   under "Processor type and features"::
180
181	CONFIG_SMP=n
182
183   (If CONFIG_SMP=y, then specify maxcpus=1 on the kernel command line
184   when loading the dump-capture kernel, see section "Load the Dump-capture
185   Kernel".)
186
1873) If one wants to build and use a relocatable kernel,
188   Enable "Build a relocatable kernel" support under "Processor type and
189   features"::
190
191	CONFIG_RELOCATABLE=y
192
1934) Use a suitable value for "Physical address where the kernel is
194   loaded" (under "Processor type and features"). This only appears when
195   "kernel crash dumps" is enabled. A suitable value depends upon
196   whether kernel is relocatable or not.
197
198   If you are using a relocatable kernel use CONFIG_PHYSICAL_START=0x100000
199   This will compile the kernel for physical address 1MB, but given the fact
200   kernel is relocatable, it can be run from any physical address hence
201   kexec boot loader will load it in memory region reserved for dump-capture
202   kernel.
203
204   Otherwise it should be the start of memory region reserved for
205   second kernel using boot parameter "crashkernel=Y@X". Here X is
206   start of memory region reserved for dump-capture kernel.
207   Generally X is 16MB (0x1000000). So you can set
208   CONFIG_PHYSICAL_START=0x1000000
209
2105) Make and install the kernel and its modules. DO NOT add this kernel
211   to the boot loader configuration files.
212
213Dump-capture kernel config options (Arch Dependent, ppc64)
214----------------------------------------------------------
215
2161) Enable "Build a kdump crash kernel" support under "Kernel" options::
217
218	CONFIG_CRASH_DUMP=y
219
2202)   Enable "Build a relocatable kernel" support::
221
222	CONFIG_RELOCATABLE=y
223
224   Make and install the kernel and its modules.
225
226Dump-capture kernel config options (Arch Dependent, ia64)
227----------------------------------------------------------
228
229- No specific options are required to create a dump-capture kernel
230  for ia64, other than those specified in the arch independent section
231  above. This means that it is possible to use the system kernel
232  as a dump-capture kernel if desired.
233
234  The crashkernel region can be automatically placed by the system
235  kernel at run time. This is done by specifying the base address as 0,
236  or omitting it all together::
237
238	crashkernel=256M@0
239
240  or::
241
242	crashkernel=256M
243
244  If the start address is specified, note that the start address of the
245  kernel will be aligned to 64Mb, so if the start address is not then
246  any space below the alignment point will be wasted.
247
248Dump-capture kernel config options (Arch Dependent, arm)
249----------------------------------------------------------
250
251-   To use a relocatable kernel,
252    Enable "AUTO_ZRELADDR" support under "Boot" options::
253
254	AUTO_ZRELADDR=y
255
256Dump-capture kernel config options (Arch Dependent, arm64)
257----------------------------------------------------------
258
259- Please note that kvm of the dump-capture kernel will not be enabled
260  on non-VHE systems even if it is configured. This is because the CPU
261  will not be reset to EL2 on panic.
262
263Extended crashkernel syntax
264===========================
265
266While the "crashkernel=size[@offset]" syntax is sufficient for most
267configurations, sometimes it's handy to have the reserved memory dependent
268on the value of System RAM -- that's mostly for distributors that pre-setup
269the kernel command line to avoid a unbootable system after some memory has
270been removed from the machine.
271
272The syntax is::
273
274    crashkernel=<range1>:<size1>[,<range2>:<size2>,...][@offset]
275    range=start-[end]
276
277For example::
278
279    crashkernel=512M-2G:64M,2G-:128M
280
281This would mean:
282
283    1) if the RAM is smaller than 512M, then don't reserve anything
284       (this is the "rescue" case)
285    2) if the RAM size is between 512M and 2G (exclusive), then reserve 64M
286    3) if the RAM size is larger than 2G, then reserve 128M
287
288
289
290Boot into System Kernel
291=======================
292
2931) Update the boot loader (such as grub, yaboot, or lilo) configuration
294   files as necessary.
295
2962) Boot the system kernel with the boot parameter "crashkernel=Y@X",
297   where Y specifies how much memory to reserve for the dump-capture kernel
298   and X specifies the beginning of this reserved memory. For example,
299   "crashkernel=64M@16M" tells the system kernel to reserve 64 MB of memory
300   starting at physical address 0x01000000 (16MB) for the dump-capture kernel.
301
302   On x86 and x86_64, use "crashkernel=64M@16M".
303
304   On ppc64, use "crashkernel=128M@32M".
305
306   On ia64, 256M@256M is a generous value that typically works.
307   The region may be automatically placed on ia64, see the
308   dump-capture kernel config option notes above.
309   If use sparse memory, the size should be rounded to GRANULE boundaries.
310
311   On s390x, typically use "crashkernel=xxM". The value of xx is dependent
312   on the memory consumption of the kdump system. In general this is not
313   dependent on the memory size of the production system.
314
315   On arm, the use of "crashkernel=Y@X" is no longer necessary; the
316   kernel will automatically locate the crash kernel image within the
317   first 512MB of RAM if X is not given.
318
319   On arm64, use "crashkernel=Y[@X]".  Note that the start address of
320   the kernel, X if explicitly specified, must be aligned to 2MiB (0x200000).
321
322Load the Dump-capture Kernel
323============================
324
325After booting to the system kernel, dump-capture kernel needs to be
326loaded.
327
328Based on the architecture and type of image (relocatable or not), one
329can choose to load the uncompressed vmlinux or compressed bzImage/vmlinuz
330of dump-capture kernel. Following is the summary.
331
332For i386 and x86_64:
333
334	- Use vmlinux if kernel is not relocatable.
335	- Use bzImage/vmlinuz if kernel is relocatable.
336
337For ppc64:
338
339	- Use vmlinux
340
341For ia64:
342
343	- Use vmlinux or vmlinuz.gz
344
345For s390x:
346
347	- Use image or bzImage
348
349For arm:
350
351	- Use zImage
352
353For arm64:
354
355	- Use vmlinux or Image
356
357If you are using an uncompressed vmlinux image then use following command
358to load dump-capture kernel::
359
360   kexec -p <dump-capture-kernel-vmlinux-image> \
361   --initrd=<initrd-for-dump-capture-kernel> --args-linux \
362   --append="root=<root-dev> <arch-specific-options>"
363
364If you are using a compressed bzImage/vmlinuz, then use following command
365to load dump-capture kernel::
366
367   kexec -p <dump-capture-kernel-bzImage> \
368   --initrd=<initrd-for-dump-capture-kernel> \
369   --append="root=<root-dev> <arch-specific-options>"
370
371If you are using a compressed zImage, then use following command
372to load dump-capture kernel::
373
374   kexec --type zImage -p <dump-capture-kernel-bzImage> \
375   --initrd=<initrd-for-dump-capture-kernel> \
376   --dtb=<dtb-for-dump-capture-kernel> \
377   --append="root=<root-dev> <arch-specific-options>"
378
379If you are using an uncompressed Image, then use following command
380to load dump-capture kernel::
381
382   kexec -p <dump-capture-kernel-Image> \
383   --initrd=<initrd-for-dump-capture-kernel> \
384   --append="root=<root-dev> <arch-specific-options>"
385
386Please note, that --args-linux does not need to be specified for ia64.
387It is planned to make this a no-op on that architecture, but for now
388it should be omitted
389
390Following are the arch specific command line options to be used while
391loading dump-capture kernel.
392
393For i386, x86_64 and ia64:
394
395	"1 irqpoll maxcpus=1 reset_devices"
396
397For ppc64:
398
399	"1 maxcpus=1 noirqdistrib reset_devices"
400
401For s390x:
402
403	"1 maxcpus=1 cgroup_disable=memory"
404
405For arm:
406
407	"1 maxcpus=1 reset_devices"
408
409For arm64:
410
411	"1 maxcpus=1 reset_devices"
412
413Notes on loading the dump-capture kernel:
414
415* By default, the ELF headers are stored in ELF64 format to support
416  systems with more than 4GB memory. On i386, kexec automatically checks if
417  the physical RAM size exceeds the 4 GB limit and if not, uses ELF32.
418  So, on non-PAE systems, ELF32 is always used.
419
420  The --elf32-core-headers option can be used to force the generation of ELF32
421  headers. This is necessary because GDB currently cannot open vmcore files
422  with ELF64 headers on 32-bit systems.
423
424* The "irqpoll" boot parameter reduces driver initialization failures
425  due to shared interrupts in the dump-capture kernel.
426
427* You must specify <root-dev> in the format corresponding to the root
428  device name in the output of mount command.
429
430* Boot parameter "1" boots the dump-capture kernel into single-user
431  mode without networking. If you want networking, use "3".
432
433* We generally don't have to bring up a SMP kernel just to capture the
434  dump. Hence generally it is useful either to build a UP dump-capture
435  kernel or specify maxcpus=1 option while loading dump-capture kernel.
436  Note, though maxcpus always works, you had better replace it with
437  nr_cpus to save memory if supported by the current ARCH, such as x86.
438
439* You should enable multi-cpu support in dump-capture kernel if you intend
440  to use multi-thread programs with it, such as parallel dump feature of
441  makedumpfile. Otherwise, the multi-thread program may have a great
442  performance degradation. To enable multi-cpu support, you should bring up an
443  SMP dump-capture kernel and specify maxcpus/nr_cpus, disable_cpu_apicid=[X]
444  options while loading it.
445
446* For s390x there are two kdump modes: If a ELF header is specified with
447  the elfcorehdr= kernel parameter, it is used by the kdump kernel as it
448  is done on all other architectures. If no elfcorehdr= kernel parameter is
449  specified, the s390x kdump kernel dynamically creates the header. The
450  second mode has the advantage that for CPU and memory hotplug, kdump has
451  not to be reloaded with kexec_load().
452
453* For s390x systems with many attached devices the "cio_ignore" kernel
454  parameter should be used for the kdump kernel in order to prevent allocation
455  of kernel memory for devices that are not relevant for kdump. The same
456  applies to systems that use SCSI/FCP devices. In that case the
457  "allow_lun_scan" zfcp module parameter should be set to zero before
458  setting FCP devices online.
459
460Kernel Panic
461============
462
463After successfully loading the dump-capture kernel as previously
464described, the system will reboot into the dump-capture kernel if a
465system crash is triggered.  Trigger points are located in panic(),
466die(), die_nmi() and in the sysrq handler (ALT-SysRq-c).
467
468The following conditions will execute a crash trigger point:
469
470If a hard lockup is detected and "NMI watchdog" is configured, the system
471will boot into the dump-capture kernel ( die_nmi() ).
472
473If die() is called, and it happens to be a thread with pid 0 or 1, or die()
474is called inside interrupt context or die() is called and panic_on_oops is set,
475the system will boot into the dump-capture kernel.
476
477On powerpc systems when a soft-reset is generated, die() is called by all cpus
478and the system will boot into the dump-capture kernel.
479
480For testing purposes, you can trigger a crash by using "ALT-SysRq-c",
481"echo c > /proc/sysrq-trigger" or write a module to force the panic.
482
483Write Out the Dump File
484=======================
485
486After the dump-capture kernel is booted, write out the dump file with
487the following command::
488
489   cp /proc/vmcore <dump-file>
490
491
492Analysis
493========
494
495Before analyzing the dump image, you should reboot into a stable kernel.
496
497You can do limited analysis using GDB on the dump file copied out of
498/proc/vmcore. Use the debug vmlinux built with -g and run the following
499command::
500
501   gdb vmlinux <dump-file>
502
503Stack trace for the task on processor 0, register display, and memory
504display work fine.
505
506Note: GDB cannot analyze core files generated in ELF64 format for x86.
507On systems with a maximum of 4GB of memory, you can generate
508ELF32-format headers using the --elf32-core-headers kernel option on the
509dump kernel.
510
511You can also use the Crash utility to analyze dump files in Kdump
512format. Crash is available on Dave Anderson's site at the following URL:
513
514   http://people.redhat.com/~anderson/
515
516Trigger Kdump on WARN()
517=======================
518
519The kernel parameter, panic_on_warn, calls panic() in all WARN() paths.  This
520will cause a kdump to occur at the panic() call.  In cases where a user wants
521to specify this during runtime, /proc/sys/kernel/panic_on_warn can be set to 1
522to achieve the same behaviour.
523
524Trigger Kdump on add_taint()
525============================
526
527The kernel parameter panic_on_taint facilitates a conditional call to panic()
528from within add_taint() whenever the value set in this bitmask matches with the
529bit flag being set by add_taint().
530This will cause a kdump to occur at the add_taint()->panic() call.
531
532Contact
533=======
534
535- Vivek Goyal (vgoyal@redhat.com)
536- Maneesh Soni (maneesh@in.ibm.com)
537
538GDB macros
539==========
540
541.. include:: gdbmacros.txt
542   :literal:
543