blob: 68aa1650973af2dba3bcf6ad75cb6a7162f6dbd1 [file] [log] [blame]
David Wilderdc851a02006-06-25 05:47:55 -07001================================================================
2Documentation for Kdump - The kexec-based Crash Dumping Solution
Vivek Goyalb089f4a2005-06-25 14:58:15 -07003================================================================
4
David Wilderdc851a02006-06-25 05:47:55 -07005This document includes overview, setup and installation, and analysis
6information.
Vivek Goyalb089f4a2005-06-25 14:58:15 -07007
David Wilderdc851a02006-06-25 05:47:55 -07008Overview
9========
Vivek Goyalb089f4a2005-06-25 14:58:15 -070010
David Wilderdc851a02006-06-25 05:47:55 -070011Kdump 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.
Vivek Goyalb089f4a2005-06-25 14:58:15 -070015
David Wilderdc851a02006-06-25 05:47:55 -070016You can use common Linux 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.
Vivek Goyalb089f4a2005-06-25 14:58:15 -070019
Hormsee8bb9e2007-01-22 20:40:49 -080020Kdump and kexec are currently supported on the x86, x86_64, ppc64 and ia64
David Wilderdc851a02006-06-25 05:47:55 -070021architectures.
Vivek Goyalb089f4a2005-06-25 14:58:15 -070022
David Wilderdc851a02006-06-25 05:47:55 -070023When 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.
Vivek Goyalb089f4a2005-06-25 14:58:15 -070028
David Wilderdc851a02006-06-25 05:47:55 -070029On 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.
Vivek Goyalb089f4a2005-06-25 14:58:15 -070032
Simon Horman30430132007-02-20 13:58:07 -080033Similarly 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
David Wilderdc851a02006-06-25 05:47:55 -070037All of the necessary information about the system kernel's core image is
38encoded in the ELF format, and stored in a reserved area of memory
39before a crash. The physical address of the start of the ELF header is
40passed to the dump-capture kernel through the elfcorehdr= boot
41parameter.
Vivek Goyalb089f4a2005-06-25 14:58:15 -070042
David Wilderdc851a02006-06-25 05:47:55 -070043With the dump-capture kernel, you can access the memory image, or "old
44memory," in two ways:
Vivek Goyalb089f4a2005-06-25 14:58:15 -070045
David Wilderdc851a02006-06-25 05:47:55 -070046- Through a /dev/oldmem device interface. A capture utility can read the
47 device file and write out the memory in raw format. This is a raw dump
48 of memory. Analysis and capture tools must be intelligent enough to
49 determine where to look for the right information.
Vivek Goyalb089f4a2005-06-25 14:58:15 -070050
David Wilderdc851a02006-06-25 05:47:55 -070051- Through /proc/vmcore. This exports the dump as an ELF-format file that
52 you can write out using file copy commands such as cp or scp. Further,
53 you can use analysis tools such as the GNU Debugger (GDB) and the Crash
54 tool to debug the dump file. This method ensures that the dump pages are
55 correctly ordered.
Vivek Goyalb089f4a2005-06-25 14:58:15 -070056
Vivek Goyalb089f4a2005-06-25 14:58:15 -070057
David Wilderdc851a02006-06-25 05:47:55 -070058Setup and Installation
59======================
Vivek Goyalb089f4a2005-06-25 14:58:15 -070060
Vivek Goyal9c61a442007-01-10 23:15:35 -080061Install kexec-tools
62-------------------
Vivek Goyalb089f4a2005-06-25 14:58:15 -070063
David Wilderdc851a02006-06-25 05:47:55 -0700641) Login as the root user.
Vivek Goyalb089f4a2005-06-25 14:58:15 -070065
David Wilderdc851a02006-06-25 05:47:55 -0700662) Download the kexec-tools user-space package from the following URL:
Vivek Goyalb089f4a2005-06-25 14:58:15 -070067
Hormsea112bd2007-01-22 20:40:48 -080068http://www.kernel.org/pub/linux/kernel/people/horms/kexec-tools/kexec-tools-testing.tar.gz
69
70This is a symlink to the latest version, which at the time of writing is
7120061214, the only release of kexec-tools-testing so far. As other versions
72are made released, the older onese will remain available at
73http://www.kernel.org/pub/linux/kernel/people/horms/kexec-tools/
Vivek Goyal9c61a442007-01-10 23:15:35 -080074
75Note: Latest kexec-tools-testing git tree is available at
76
77git://git.kernel.org/pub/scm/linux/kernel/git/horms/kexec-tools-testing.git
78or
79http://www.kernel.org/git/?p=linux/kernel/git/horms/kexec-tools-testing.git;a=summary
Vivek Goyalb089f4a2005-06-25 14:58:15 -070080
David Wilderdc851a02006-06-25 05:47:55 -0700813) Unpack the tarball with the tar command, as follows:
82
Hormsea112bd2007-01-22 20:40:48 -080083 tar xvpzf kexec-tools-testing.tar.gz
David Wilderdc851a02006-06-25 05:47:55 -070084
Hormsea112bd2007-01-22 20:40:48 -0800854) Change to the kexec-tools directory, as follows:
David Wilderdc851a02006-06-25 05:47:55 -070086
Hormsea112bd2007-01-22 20:40:48 -080087 cd kexec-tools-testing-VERSION
David Wilderdc851a02006-06-25 05:47:55 -070088
Vivek Goyal9c61a442007-01-10 23:15:35 -0800895) Configure the package, as follows:
David Wilderdc851a02006-06-25 05:47:55 -070090
91 ./configure
92
Vivek Goyal9c61a442007-01-10 23:15:35 -0800936) Compile the package, as follows:
David Wilderdc851a02006-06-25 05:47:55 -070094
95 make
96
Vivek Goyal9c61a442007-01-10 23:15:35 -0800977) Install the package, as follows:
David Wilderdc851a02006-06-25 05:47:55 -070098
99 make install
100
101
Vivek Goyal9c61a442007-01-10 23:15:35 -0800102Build the system and dump-capture kernels
103-----------------------------------------
104There are two possible methods of using Kdump.
David Wilderdc851a02006-06-25 05:47:55 -0700105
Vivek Goyal9c61a442007-01-10 23:15:35 -08001061) Build a separate custom dump-capture kernel for capturing the
107 kernel core dump.
David Wilderdc851a02006-06-25 05:47:55 -0700108
Vivek Goyal9c61a442007-01-10 23:15:35 -08001092) Or use the system kernel binary itself as dump-capture kernel and there is
110 no need to build a separate dump-capture kernel. This is possible
111 only with the architecutres which support a relocatable kernel. As
112 of today i386 and ia64 architectures support relocatable kernel.
113
114Building a relocatable kernel is advantageous from the point of view that
115one does not have to build a second kernel for capturing the dump. But
116at the same time one might want to build a custom dump capture kernel
117suitable to his needs.
118
119Following are the configuration setting required for system and
120dump-capture kernels for enabling kdump support.
121
122System kernel config options
123----------------------------
David Wilderdc851a02006-06-25 05:47:55 -0700124
1251) Enable "kexec system call" in "Processor type and features."
126
127 CONFIG_KEXEC=y
128
1292) Enable "sysfs file system support" in "Filesystem" -> "Pseudo
130 filesystems." This is usually enabled by default.
131
132 CONFIG_SYSFS=y
133
134 Note that "sysfs file system support" might not appear in the "Pseudo
135 filesystems" menu if "Configure standard kernel features (for small
136 systems)" is not enabled in "General Setup." In this case, check the
137 .config file itself to ensure that sysfs is turned on, as follows:
138
139 grep 'CONFIG_SYSFS' .config
140
1413) Enable "Compile the kernel with debug info" in "Kernel hacking."
142
143 CONFIG_DEBUG_INFO=Y
144
145 This causes the kernel to be built with debug symbols. The dump
146 analysis tools require a vmlinux with debug symbols in order to read
147 and analyze a dump file.
148
Vivek Goyal9c61a442007-01-10 23:15:35 -0800149Dump-capture kernel config options (Arch Independent)
150-----------------------------------------------------
David Wilderdc851a02006-06-25 05:47:55 -0700151
Vivek Goyal9c61a442007-01-10 23:15:35 -08001521) Enable "kernel crash dumps" support under "Processor type and
153 features":
David Wilderdc851a02006-06-25 05:47:55 -0700154
Vivek Goyal9c61a442007-01-10 23:15:35 -0800155 CONFIG_CRASH_DUMP=y
David Wilderdc851a02006-06-25 05:47:55 -0700156
Vivek Goyal9c61a442007-01-10 23:15:35 -08001572) Enable "/proc/vmcore support" under "Filesystems" -> "Pseudo filesystems".
David Wilderdc851a02006-06-25 05:47:55 -0700158
Vivek Goyal9c61a442007-01-10 23:15:35 -0800159 CONFIG_PROC_VMCORE=y
160 (CONFIG_PROC_VMCORE is set by default when CONFIG_CRASH_DUMP is selected.)
David Wilderdc851a02006-06-25 05:47:55 -0700161
Bernhard Walle8bc9d422007-10-16 23:31:21 -0700162Dump-capture kernel config options (Arch Dependent, i386 and x86_64)
163--------------------------------------------------------------------
164
1651) On i386, enable high memory support under "Processor type and
David Wilderdc851a02006-06-25 05:47:55 -0700166 features":
167
168 CONFIG_HIGHMEM64G=y
169 or
170 CONFIG_HIGHMEM4G
171
Bernhard Walle8bc9d422007-10-16 23:31:21 -07001722) On i386 and x86_64, disable symmetric multi-processing support
David Wilderdc851a02006-06-25 05:47:55 -0700173 under "Processor type and features":
174
175 CONFIG_SMP=n
Vivek Goyal9c61a442007-01-10 23:15:35 -0800176
David Wilderdc851a02006-06-25 05:47:55 -0700177 (If CONFIG_SMP=y, then specify maxcpus=1 on the kernel command line
178 when loading the dump-capture kernel, see section "Load the Dump-capture
179 Kernel".)
180
Vivek Goyal9c61a442007-01-10 23:15:35 -08001813) If one wants to build and use a relocatable kernel,
182 Enable "Build a relocatable kernel" support under "Processor type and
183 features"
David Wilderdc851a02006-06-25 05:47:55 -0700184
Vivek Goyal9c61a442007-01-10 23:15:35 -0800185 CONFIG_RELOCATABLE=y
David Wilderdc851a02006-06-25 05:47:55 -0700186
Vivek Goyal9c61a442007-01-10 23:15:35 -08001874) Use a suitable value for "Physical address where the kernel is
188 loaded" (under "Processor type and features"). This only appears when
189 "kernel crash dumps" is enabled. A suitable value depends upon
190 whether kernel is relocatable or not.
David Wilderdc851a02006-06-25 05:47:55 -0700191
Vivek Goyal9c61a442007-01-10 23:15:35 -0800192 If you are using a relocatable kernel use CONFIG_PHYSICAL_START=0x100000
193 This will compile the kernel for physical address 1MB, but given the fact
194 kernel is relocatable, it can be run from any physical address hence
195 kexec boot loader will load it in memory region reserved for dump-capture
196 kernel.
David Wilderdc851a02006-06-25 05:47:55 -0700197
Vivek Goyal9c61a442007-01-10 23:15:35 -0800198 Otherwise it should be the start of memory region reserved for
199 second kernel using boot parameter "crashkernel=Y@X". Here X is
200 start of memory region reserved for dump-capture kernel.
201 Generally X is 16MB (0x1000000). So you can set
202 CONFIG_PHYSICAL_START=0x1000000
203
2045) Make and install the kernel and its modules. DO NOT add this kernel
205 to the boot loader configuration files.
206
Vivek Goyal9c61a442007-01-10 23:15:35 -0800207Dump-capture kernel config options (Arch Dependent, ppc64)
208----------------------------------------------------------
209
Simon Horman30430132007-02-20 13:58:07 -0800210* Make and install the kernel and its modules. DO NOT add this kernel
Vivek Goyal9c61a442007-01-10 23:15:35 -0800211 to the boot loader configuration files.
212
213Dump-capture kernel config options (Arch Dependent, ia64)
214----------------------------------------------------------
Hormsee8bb9e2007-01-22 20:40:49 -0800215
216- No specific options are required to create a dump-capture kernel
217 for ia64, other than those specified in the arch idependent section
218 above. This means that it is possible to use the system kernel
219 as a dump-capture kernel if desired.
220
221 The crashkernel region can be automatically placed by the system
222 kernel at run time. This is done by specifying the base address as 0,
223 or omitting it all together.
224
225 crashkernel=256M@0
226 or
227 crashkernel=256M
228
229 If the start address is specified, note that the start address of the
230 kernel will be aligned to 64Mb, so if the start address is not then
231 any space below the alignment point will be wasted.
Vivek Goyal9c61a442007-01-10 23:15:35 -0800232
233
234Boot into System Kernel
235=======================
236
Simon Horman30430132007-02-20 13:58:07 -08002371) Update the boot loader (such as grub, yaboot, or lilo) configuration
238 files as necessary.
Vivek Goyal9c61a442007-01-10 23:15:35 -0800239
2402) Boot the system kernel with the boot parameter "crashkernel=Y@X",
241 where Y specifies how much memory to reserve for the dump-capture kernel
242 and X specifies the beginning of this reserved memory. For example,
243 "crashkernel=64M@16M" tells the system kernel to reserve 64 MB of memory
244 starting at physical address 0x01000000 (16MB) for the dump-capture kernel.
245
246 On x86 and x86_64, use "crashkernel=64M@16M".
247
248 On ppc64, use "crashkernel=128M@32M".
David Wilderdc851a02006-06-25 05:47:55 -0700249
Hormsee8bb9e2007-01-22 20:40:49 -0800250 On ia64, 256M@256M is a generous value that typically works.
251 The region may be automatically placed on ia64, see the
252 dump-capture kernel config option notes above.
253
David Wilderdc851a02006-06-25 05:47:55 -0700254Load the Dump-capture Kernel
255============================
256
Vivek Goyal9c61a442007-01-10 23:15:35 -0800257After booting to the system kernel, dump-capture kernel needs to be
258loaded.
David Wilderdc851a02006-06-25 05:47:55 -0700259
Vivek Goyal9c61a442007-01-10 23:15:35 -0800260Based on the architecture and type of image (relocatable or not), one
261can choose to load the uncompressed vmlinux or compressed bzImage/vmlinuz
262of dump-capture kernel. Following is the summary.
263
Bernhard Walle8bc9d422007-10-16 23:31:21 -0700264For i386 and x86_64:
Vivek Goyal9c61a442007-01-10 23:15:35 -0800265 - Use vmlinux if kernel is not relocatable.
266 - Use bzImage/vmlinuz if kernel is relocatable.
Vivek Goyal9c61a442007-01-10 23:15:35 -0800267For ppc64:
268 - Use vmlinux
269For ia64:
Hormsee8bb9e2007-01-22 20:40:49 -0800270 - Use vmlinux or vmlinuz.gz
271
Vivek Goyal9c61a442007-01-10 23:15:35 -0800272
273If you are using a uncompressed vmlinux image then use following command
274to load dump-capture kernel.
275
276 kexec -p <dump-capture-kernel-vmlinux-image> \
David Wilderdc851a02006-06-25 05:47:55 -0700277 --initrd=<initrd-for-dump-capture-kernel> --args-linux \
Vivek Goyal9c61a442007-01-10 23:15:35 -0800278 --append="root=<root-dev> <arch-specific-options>"
279
280If you are using a compressed bzImage/vmlinuz, then use following command
281to load dump-capture kernel.
282
283 kexec -p <dump-capture-kernel-bzImage> \
284 --initrd=<initrd-for-dump-capture-kernel> \
285 --append="root=<root-dev> <arch-specific-options>"
286
Hormsee8bb9e2007-01-22 20:40:49 -0800287Please note, that --args-linux does not need to be specified for ia64.
288It is planned to make this a no-op on that architecture, but for now
289it should be omitted
290
Vivek Goyal9c61a442007-01-10 23:15:35 -0800291Following are the arch specific command line options to be used while
292loading dump-capture kernel.
293
Hormsee8bb9e2007-01-22 20:40:49 -0800294For i386, x86_64 and ia64:
Horms473e66f2007-02-12 00:52:18 -0800295 "1 irqpoll maxcpus=1"
Vivek Goyal9c61a442007-01-10 23:15:35 -0800296
297For ppc64:
Horms473e66f2007-02-12 00:52:18 -0800298 "1 maxcpus=1 noirqdistrib"
Vivek Goyal9c61a442007-01-10 23:15:35 -0800299
David Wilderdc851a02006-06-25 05:47:55 -0700300
301Notes on loading the dump-capture kernel:
302
David Wilderdc851a02006-06-25 05:47:55 -0700303* By default, the ELF headers are stored in ELF64 format to support
304 systems with more than 4GB memory. The --elf32-core-headers option can
305 be used to force the generation of ELF32 headers. This is necessary
306 because GDB currently cannot open vmcore files with ELF64 headers on
307 32-bit systems. ELF32 headers can be used on non-PAE systems (that is,
308 less than 4GB of memory).
309
310* The "irqpoll" boot parameter reduces driver initialization failures
311 due to shared interrupts in the dump-capture kernel.
312
313* You must specify <root-dev> in the format corresponding to the root
314 device name in the output of mount command.
315
Horms473e66f2007-02-12 00:52:18 -0800316* Boot parameter "1" boots the dump-capture kernel into single-user
317 mode without networking. If you want networking, use "3".
David Wilderdc851a02006-06-25 05:47:55 -0700318
Vivek Goyal9c61a442007-01-10 23:15:35 -0800319* We generally don' have to bring up a SMP kernel just to capture the
320 dump. Hence generally it is useful either to build a UP dump-capture
321 kernel or specify maxcpus=1 option while loading dump-capture kernel.
David Wilderdc851a02006-06-25 05:47:55 -0700322
323Kernel Panic
324============
325
326After successfully loading the dump-capture kernel as previously
327described, the system will reboot into the dump-capture kernel if a
328system crash is triggered. Trigger points are located in panic(),
329die(), die_nmi() and in the sysrq handler (ALT-SysRq-c).
330
331The following conditions will execute a crash trigger point:
332
333If a hard lockup is detected and "NMI watchdog" is configured, the system
334will boot into the dump-capture kernel ( die_nmi() ).
335
336If die() is called, and it happens to be a thread with pid 0 or 1, or die()
337is called inside interrupt context or die() is called and panic_on_oops is set,
338the system will boot into the dump-capture kernel.
339
Simon Horman30430132007-02-20 13:58:07 -0800340On powererpc systems when a soft-reset is generated, die() is called by all cpus
341and the system will boot into the dump-capture kernel.
David Wilderdc851a02006-06-25 05:47:55 -0700342
343For testing purposes, you can trigger a crash by using "ALT-SysRq-c",
Simon Horman30430132007-02-20 13:58:07 -0800344"echo c > /proc/sysrq-trigger" or write a module to force the panic.
David Wilderdc851a02006-06-25 05:47:55 -0700345
346Write Out the Dump File
347=======================
348
349After the dump-capture kernel is booted, write out the dump file with
350the following command:
Vivek Goyalb089f4a2005-06-25 14:58:15 -0700351
352 cp /proc/vmcore <dump-file>
353
David Wilderdc851a02006-06-25 05:47:55 -0700354You can also access dumped memory as a /dev/oldmem device for a linear
355and raw view. To create the device, use the following command:
Vivek Goyalb089f4a2005-06-25 14:58:15 -0700356
David Wilderdc851a02006-06-25 05:47:55 -0700357 mknod /dev/oldmem c 1 12
Vivek Goyalb089f4a2005-06-25 14:58:15 -0700358
David Wilderdc851a02006-06-25 05:47:55 -0700359Use the dd command with suitable options for count, bs, and skip to
360access specific portions of the dump.
Vivek Goyalb089f4a2005-06-25 14:58:15 -0700361
David Wilderdc851a02006-06-25 05:47:55 -0700362To see the entire memory, use the following command:
363
364 dd if=/dev/oldmem of=oldmem.001
Vivek Goyalb089f4a2005-06-25 14:58:15 -0700365
Maneesh Sonia7e670d2006-01-09 20:51:53 -0800366
David Wilderdc851a02006-06-25 05:47:55 -0700367Analysis
Vivek Goyalb089f4a2005-06-25 14:58:15 -0700368========
Vivek Goyalb089f4a2005-06-25 14:58:15 -0700369
David Wilderdc851a02006-06-25 05:47:55 -0700370Before analyzing the dump image, you should reboot into a stable kernel.
Vivek Goyalb089f4a2005-06-25 14:58:15 -0700371
David Wilderdc851a02006-06-25 05:47:55 -0700372You can do limited analysis using GDB on the dump file copied out of
373/proc/vmcore. Use the debug vmlinux built with -g and run the following
374command:
Vivek Goyalb089f4a2005-06-25 14:58:15 -0700375
David Wilderdc851a02006-06-25 05:47:55 -0700376 gdb vmlinux <dump-file>
Vivek Goyalb089f4a2005-06-25 14:58:15 -0700377
David Wilderdc851a02006-06-25 05:47:55 -0700378Stack trace for the task on processor 0, register display, and memory
379display work fine.
380
381Note: GDB cannot analyze core files generated in ELF64 format for x86.
382On systems with a maximum of 4GB of memory, you can generate
383ELF32-format headers using the --elf32-core-headers kernel option on the
384dump kernel.
385
386You can also use the Crash utility to analyze dump files in Kdump
387format. Crash is available on Dave Anderson's site at the following URL:
388
389 http://people.redhat.com/~anderson/
Maneesh Sonia7e670d2006-01-09 20:51:53 -0800390
391
David Wilderdc851a02006-06-25 05:47:55 -0700392To Do
393=====
394
Simon Horman30430132007-02-20 13:58:07 -08003951) Provide relocatable kernels for all architectures to help in maintaining
396 multiple kernels for crash_dump, and the same kernel as the system kernel
397 can be used to capture the dump.
Maneesh Sonia7e670d2006-01-09 20:51:53 -0800398
Vivek Goyalb089f4a2005-06-25 14:58:15 -0700399
David Wilderdc851a02006-06-25 05:47:55 -0700400Contact
Vivek Goyalb089f4a2005-06-25 14:58:15 -0700401=======
David Wilderdc851a02006-06-25 05:47:55 -0700402
Vivek Goyalb089f4a2005-06-25 14:58:15 -0700403Vivek Goyal (vgoyal@in.ibm.com)
Vivek Goyald58831e2005-06-25 14:58:17 -0700404Maneesh Soni (maneesh@in.ibm.com)
David Wilderdc851a02006-06-25 05:47:55 -0700405
406
407Trademark
408=========
409
410Linux is a trademark of Linus Torvalds in the United States, other
411countries, or both.