Commit | Line | Data |
---|---|---|
dc851a0f DW |
1 | ================================================================ |
2 | Documentation for Kdump - The kexec-based Crash Dumping Solution | |
b089f4a6 VG |
3 | ================================================================ |
4 | ||
dc851a0f DW |
5 | This document includes overview, setup and installation, and analysis |
6 | information. | |
b089f4a6 | 7 | |
dc851a0f DW |
8 | Overview |
9 | ======== | |
b089f4a6 | 10 | |
dc851a0f DW |
11 | Kdump uses kexec to quickly boot to a dump-capture kernel whenever a |
12 | dump of the system kernel's memory needs to be taken (for example, when | |
13 | the system panics). The system kernel's memory image is preserved across | |
14 | the reboot and is accessible to the dump-capture kernel. | |
b089f4a6 | 15 | |
dc851a0f DW |
16 | You can use common Linux commands, such as cp and scp, to copy the |
17 | memory image to a dump file on the local disk, or across the network to | |
18 | a remote system. | |
b089f4a6 | 19 | |
ee8bb9ea | 20 | Kdump and kexec are currently supported on the x86, x86_64, ppc64 and ia64 |
dc851a0f | 21 | architectures. |
b089f4a6 | 22 | |
dc851a0f DW |
23 | When the system kernel boots, it reserves a small section of memory for |
24 | the dump-capture kernel. This ensures that ongoing Direct Memory Access | |
25 | (DMA) from the system kernel does not corrupt the dump-capture kernel. | |
26 | The kexec -p command loads the dump-capture kernel into this reserved | |
27 | memory. | |
b089f4a6 | 28 | |
dc851a0f DW |
29 | On x86 machines, the first 640 KB of physical memory is needed to boot, |
30 | regardless of where the kernel loads. Therefore, kexec backs up this | |
31 | region just before rebooting into the dump-capture kernel. | |
b089f4a6 | 32 | |
30430134 SH |
33 | Similarly on PPC64 machines first 32KB of physical memory is needed for |
34 | booting regardless of where the kernel is loaded and to support 64K page | |
35 | size kexec backs up the first 64KB memory. | |
36 | ||
dc851a0f DW |
37 | All of the necessary information about the system kernel's core image is |
38 | encoded in the ELF format, and stored in a reserved area of memory | |
39 | before a crash. The physical address of the start of the ELF header is | |
40 | passed to the dump-capture kernel through the elfcorehdr= boot | |
41 | parameter. | |
42 | ||
43 | With the dump-capture kernel, you can access the memory image, or "old | |
44 | memory," in two ways: | |
45 | ||
46 | - 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. | |
50 | ||
51 | - 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. | |
56 | ||
57 | ||
58 | Setup and Installation | |
59 | ====================== | |
60 | ||
9c61a446 VG |
61 | Install kexec-tools |
62 | ------------------- | |
dc851a0f DW |
63 | |
64 | 1) Login as the root user. | |
65 | ||
66 | 2) Download the kexec-tools user-space package from the following URL: | |
67 | ||
ea112bd5 H |
68 | http://www.kernel.org/pub/linux/kernel/people/horms/kexec-tools/kexec-tools-testing.tar.gz |
69 | ||
70 | This is a symlink to the latest version, which at the time of writing is | |
71 | 20061214, the only release of kexec-tools-testing so far. As other versions | |
72 | are made released, the older onese will remain available at | |
73 | http://www.kernel.org/pub/linux/kernel/people/horms/kexec-tools/ | |
dc851a0f | 74 | |
9c61a446 | 75 | Note: Latest kexec-tools-testing git tree is available at |
dc851a0f | 76 | |
9c61a446 VG |
77 | git://git.kernel.org/pub/scm/linux/kernel/git/horms/kexec-tools-testing.git |
78 | or | |
79 | http://www.kernel.org/git/?p=linux/kernel/git/horms/kexec-tools-testing.git;a=summary | |
dc851a0f | 80 | |
9c61a446 | 81 | 3) Unpack the tarball with the tar command, as follows: |
dc851a0f | 82 | |
ea112bd5 | 83 | tar xvpzf kexec-tools-testing.tar.gz |
dc851a0f | 84 | |
ea112bd5 | 85 | 4) Change to the kexec-tools directory, as follows: |
dc851a0f | 86 | |
ea112bd5 | 87 | cd kexec-tools-testing-VERSION |
dc851a0f | 88 | |
9c61a446 | 89 | 5) Configure the package, as follows: |
dc851a0f DW |
90 | |
91 | ./configure | |
92 | ||
9c61a446 | 93 | 6) Compile the package, as follows: |
dc851a0f DW |
94 | |
95 | make | |
96 | ||
9c61a446 | 97 | 7) Install the package, as follows: |
dc851a0f DW |
98 | |
99 | make install | |
100 | ||
101 | ||
9c61a446 VG |
102 | Build the system and dump-capture kernels |
103 | ----------------------------------------- | |
104 | There are two possible methods of using Kdump. | |
105 | ||
106 | 1) Build a separate custom dump-capture kernel for capturing the | |
107 | kernel core dump. | |
108 | ||
109 | 2) 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 | ||
114 | Building a relocatable kernel is advantageous from the point of view that | |
115 | one does not have to build a second kernel for capturing the dump. But | |
116 | at the same time one might want to build a custom dump capture kernel | |
117 | suitable to his needs. | |
dc851a0f | 118 | |
9c61a446 VG |
119 | Following are the configuration setting required for system and |
120 | dump-capture kernels for enabling kdump support. | |
dc851a0f | 121 | |
9c61a446 VG |
122 | System kernel config options |
123 | ---------------------------- | |
dc851a0f DW |
124 | |
125 | 1) Enable "kexec system call" in "Processor type and features." | |
126 | ||
127 | CONFIG_KEXEC=y | |
128 | ||
129 | 2) 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 | ||
141 | 3) 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 | ||
9c61a446 VG |
149 | Dump-capture kernel config options (Arch Independent) |
150 | ----------------------------------------------------- | |
dc851a0f | 151 | |
9c61a446 VG |
152 | 1) Enable "kernel crash dumps" support under "Processor type and |
153 | features": | |
dc851a0f | 154 | |
9c61a446 | 155 | CONFIG_CRASH_DUMP=y |
dc851a0f | 156 | |
9c61a446 | 157 | 2) Enable "/proc/vmcore support" under "Filesystems" -> "Pseudo filesystems". |
b089f4a6 | 158 | |
9c61a446 VG |
159 | CONFIG_PROC_VMCORE=y |
160 | (CONFIG_PROC_VMCORE is set by default when CONFIG_CRASH_DUMP is selected.) | |
dc851a0f | 161 | |
9c61a446 VG |
162 | Dump-capture kernel config options (Arch Dependent, i386) |
163 | -------------------------------------------------------- | |
164 | 1) On x86, enable high memory support under "Processor type and | |
dc851a0f DW |
165 | features": |
166 | ||
167 | CONFIG_HIGHMEM64G=y | |
168 | or | |
169 | CONFIG_HIGHMEM4G | |
170 | ||
9c61a446 | 171 | 2) On x86 and x86_64, disable symmetric multi-processing support |
dc851a0f DW |
172 | under "Processor type and features": |
173 | ||
174 | CONFIG_SMP=n | |
9c61a446 | 175 | |
dc851a0f DW |
176 | (If CONFIG_SMP=y, then specify maxcpus=1 on the kernel command line |
177 | when loading the dump-capture kernel, see section "Load the Dump-capture | |
178 | Kernel".) | |
179 | ||
9c61a446 VG |
180 | 3) If one wants to build and use a relocatable kernel, |
181 | Enable "Build a relocatable kernel" support under "Processor type and | |
182 | features" | |
dc851a0f | 183 | |
9c61a446 | 184 | CONFIG_RELOCATABLE=y |
dc851a0f | 185 | |
9c61a446 VG |
186 | 4) Use a suitable value for "Physical address where the kernel is |
187 | loaded" (under "Processor type and features"). This only appears when | |
188 | "kernel crash dumps" is enabled. A suitable value depends upon | |
189 | whether kernel is relocatable or not. | |
190 | ||
191 | If you are using a relocatable kernel use CONFIG_PHYSICAL_START=0x100000 | |
192 | This will compile the kernel for physical address 1MB, but given the fact | |
193 | kernel is relocatable, it can be run from any physical address hence | |
194 | kexec boot loader will load it in memory region reserved for dump-capture | |
195 | kernel. | |
196 | ||
197 | Otherwise it should be the start of memory region reserved for | |
198 | second kernel using boot parameter "crashkernel=Y@X". Here X is | |
199 | start of memory region reserved for dump-capture kernel. | |
200 | Generally X is 16MB (0x1000000). So you can set | |
201 | CONFIG_PHYSICAL_START=0x1000000 | |
202 | ||
203 | 5) Make and install the kernel and its modules. DO NOT add this kernel | |
204 | to the boot loader configuration files. | |
dc851a0f | 205 | |
9c61a446 VG |
206 | Dump-capture kernel config options (Arch Dependent, x86_64) |
207 | ---------------------------------------------------------- | |
208 | 1) On x86 and x86_64, disable symmetric multi-processing support | |
209 | under "Processor type and features": | |
210 | ||
211 | CONFIG_SMP=n | |
212 | ||
213 | (If CONFIG_SMP=y, then specify maxcpus=1 on the kernel command line | |
214 | when loading the dump-capture kernel, see section "Load the Dump-capture | |
215 | Kernel".) | |
dc851a0f | 216 | |
9c61a446 | 217 | 2) Use a suitable value for "Physical address where the kernel is |
dc851a0f DW |
218 | loaded" (under "Processor type and features"). This only appears when |
219 | "kernel crash dumps" is enabled. By default this value is 0x1000000 | |
220 | (16MB). It should be the same as X in the "crashkernel=Y@X" boot | |
9c61a446 | 221 | parameter. |
dc851a0f | 222 | |
9c61a446 | 223 | For x86_64, normally "CONFIG_PHYSICAL_START=0x1000000". |
dc851a0f | 224 | |
9c61a446 VG |
225 | 3) Make and install the kernel and its modules. DO NOT add this kernel |
226 | to the boot loader configuration files. | |
dc851a0f | 227 | |
9c61a446 VG |
228 | Dump-capture kernel config options (Arch Dependent, ppc64) |
229 | ---------------------------------------------------------- | |
dc851a0f | 230 | |
30430134 | 231 | * Make and install the kernel and its modules. DO NOT add this kernel |
dc851a0f DW |
232 | to the boot loader configuration files. |
233 | ||
9c61a446 VG |
234 | Dump-capture kernel config options (Arch Dependent, ia64) |
235 | ---------------------------------------------------------- | |
ee8bb9ea H |
236 | |
237 | - No specific options are required to create a dump-capture kernel | |
238 | for ia64, other than those specified in the arch idependent section | |
239 | above. This means that it is possible to use the system kernel | |
240 | as a dump-capture kernel if desired. | |
241 | ||
242 | The crashkernel region can be automatically placed by the system | |
243 | kernel at run time. This is done by specifying the base address as 0, | |
244 | or omitting it all together. | |
245 | ||
246 | crashkernel=256M@0 | |
247 | or | |
248 | crashkernel=256M | |
249 | ||
250 | If the start address is specified, note that the start address of the | |
251 | kernel will be aligned to 64Mb, so if the start address is not then | |
252 | any space below the alignment point will be wasted. | |
9c61a446 VG |
253 | |
254 | ||
255 | Boot into System Kernel | |
256 | ======================= | |
257 | ||
30430134 SH |
258 | 1) Update the boot loader (such as grub, yaboot, or lilo) configuration |
259 | files as necessary. | |
9c61a446 VG |
260 | |
261 | 2) Boot the system kernel with the boot parameter "crashkernel=Y@X", | |
262 | where Y specifies how much memory to reserve for the dump-capture kernel | |
263 | and X specifies the beginning of this reserved memory. For example, | |
264 | "crashkernel=64M@16M" tells the system kernel to reserve 64 MB of memory | |
265 | starting at physical address 0x01000000 (16MB) for the dump-capture kernel. | |
266 | ||
267 | On x86 and x86_64, use "crashkernel=64M@16M". | |
268 | ||
269 | On ppc64, use "crashkernel=128M@32M". | |
dc851a0f | 270 | |
ee8bb9ea H |
271 | On ia64, 256M@256M is a generous value that typically works. |
272 | The region may be automatically placed on ia64, see the | |
273 | dump-capture kernel config option notes above. | |
274 | ||
dc851a0f DW |
275 | Load the Dump-capture Kernel |
276 | ============================ | |
277 | ||
9c61a446 VG |
278 | After booting to the system kernel, dump-capture kernel needs to be |
279 | loaded. | |
280 | ||
281 | Based on the architecture and type of image (relocatable or not), one | |
282 | can choose to load the uncompressed vmlinux or compressed bzImage/vmlinuz | |
283 | of dump-capture kernel. Following is the summary. | |
284 | ||
285 | For i386: | |
286 | - Use vmlinux if kernel is not relocatable. | |
287 | - Use bzImage/vmlinuz if kernel is relocatable. | |
288 | For x86_64: | |
289 | - Use vmlinux | |
290 | For ppc64: | |
291 | - Use vmlinux | |
292 | For ia64: | |
ee8bb9ea H |
293 | - Use vmlinux or vmlinuz.gz |
294 | ||
9c61a446 VG |
295 | |
296 | If you are using a uncompressed vmlinux image then use following command | |
297 | to load dump-capture kernel. | |
dc851a0f | 298 | |
9c61a446 | 299 | kexec -p <dump-capture-kernel-vmlinux-image> \ |
dc851a0f | 300 | --initrd=<initrd-for-dump-capture-kernel> --args-linux \ |
9c61a446 | 301 | --append="root=<root-dev> <arch-specific-options>" |
dc851a0f | 302 | |
9c61a446 VG |
303 | If you are using a compressed bzImage/vmlinuz, then use following command |
304 | to load dump-capture kernel. | |
dc851a0f | 305 | |
9c61a446 VG |
306 | kexec -p <dump-capture-kernel-bzImage> \ |
307 | --initrd=<initrd-for-dump-capture-kernel> \ | |
308 | --append="root=<root-dev> <arch-specific-options>" | |
309 | ||
ee8bb9ea H |
310 | Please note, that --args-linux does not need to be specified for ia64. |
311 | It is planned to make this a no-op on that architecture, but for now | |
312 | it should be omitted | |
313 | ||
9c61a446 VG |
314 | Following are the arch specific command line options to be used while |
315 | loading dump-capture kernel. | |
316 | ||
ee8bb9ea | 317 | For i386, x86_64 and ia64: |
473e66fd | 318 | "1 irqpoll maxcpus=1" |
9c61a446 VG |
319 | |
320 | For ppc64: | |
473e66fd | 321 | "1 maxcpus=1 noirqdistrib" |
dc851a0f | 322 | |
9c61a446 VG |
323 | |
324 | Notes on loading the dump-capture kernel: | |
dc851a0f DW |
325 | |
326 | * By default, the ELF headers are stored in ELF64 format to support | |
327 | systems with more than 4GB memory. The --elf32-core-headers option can | |
328 | be used to force the generation of ELF32 headers. This is necessary | |
329 | because GDB currently cannot open vmcore files with ELF64 headers on | |
330 | 32-bit systems. ELF32 headers can be used on non-PAE systems (that is, | |
331 | less than 4GB of memory). | |
332 | ||
333 | * The "irqpoll" boot parameter reduces driver initialization failures | |
334 | due to shared interrupts in the dump-capture kernel. | |
335 | ||
336 | * You must specify <root-dev> in the format corresponding to the root | |
337 | device name in the output of mount command. | |
338 | ||
473e66fd H |
339 | * Boot parameter "1" boots the dump-capture kernel into single-user |
340 | mode without networking. If you want networking, use "3". | |
dc851a0f | 341 | |
9c61a446 VG |
342 | * We generally don' have to bring up a SMP kernel just to capture the |
343 | dump. Hence generally it is useful either to build a UP dump-capture | |
344 | kernel or specify maxcpus=1 option while loading dump-capture kernel. | |
dc851a0f DW |
345 | |
346 | Kernel Panic | |
347 | ============ | |
348 | ||
349 | After successfully loading the dump-capture kernel as previously | |
350 | described, the system will reboot into the dump-capture kernel if a | |
351 | system crash is triggered. Trigger points are located in panic(), | |
352 | die(), die_nmi() and in the sysrq handler (ALT-SysRq-c). | |
353 | ||
354 | The following conditions will execute a crash trigger point: | |
355 | ||
356 | If a hard lockup is detected and "NMI watchdog" is configured, the system | |
357 | will boot into the dump-capture kernel ( die_nmi() ). | |
358 | ||
359 | If die() is called, and it happens to be a thread with pid 0 or 1, or die() | |
360 | is called inside interrupt context or die() is called and panic_on_oops is set, | |
361 | the system will boot into the dump-capture kernel. | |
362 | ||
30430134 SH |
363 | On powererpc systems when a soft-reset is generated, die() is called by all cpus |
364 | and the system will boot into the dump-capture kernel. | |
dc851a0f DW |
365 | |
366 | For testing purposes, you can trigger a crash by using "ALT-SysRq-c", | |
30430134 | 367 | "echo c > /proc/sysrq-trigger" or write a module to force the panic. |
dc851a0f DW |
368 | |
369 | Write Out the Dump File | |
370 | ======================= | |
371 | ||
372 | After the dump-capture kernel is booted, write out the dump file with | |
373 | the following command: | |
b089f4a6 VG |
374 | |
375 | cp /proc/vmcore <dump-file> | |
376 | ||
dc851a0f DW |
377 | You can also access dumped memory as a /dev/oldmem device for a linear |
378 | and raw view. To create the device, use the following command: | |
b089f4a6 | 379 | |
dc851a0f | 380 | mknod /dev/oldmem c 1 12 |
b089f4a6 | 381 | |
dc851a0f DW |
382 | Use the dd command with suitable options for count, bs, and skip to |
383 | access specific portions of the dump. | |
b089f4a6 | 384 | |
dc851a0f | 385 | To see the entire memory, use the following command: |
b089f4a6 | 386 | |
dc851a0f | 387 | dd if=/dev/oldmem of=oldmem.001 |
a7e670d8 | 388 | |
dc851a0f DW |
389 | |
390 | Analysis | |
b089f4a6 | 391 | ======== |
b089f4a6 | 392 | |
dc851a0f DW |
393 | Before analyzing the dump image, you should reboot into a stable kernel. |
394 | ||
395 | You can do limited analysis using GDB on the dump file copied out of | |
396 | /proc/vmcore. Use the debug vmlinux built with -g and run the following | |
397 | command: | |
398 | ||
399 | gdb vmlinux <dump-file> | |
b089f4a6 | 400 | |
dc851a0f DW |
401 | Stack trace for the task on processor 0, register display, and memory |
402 | display work fine. | |
b089f4a6 | 403 | |
dc851a0f DW |
404 | Note: GDB cannot analyze core files generated in ELF64 format for x86. |
405 | On systems with a maximum of 4GB of memory, you can generate | |
406 | ELF32-format headers using the --elf32-core-headers kernel option on the | |
407 | dump kernel. | |
b089f4a6 | 408 | |
dc851a0f DW |
409 | You can also use the Crash utility to analyze dump files in Kdump |
410 | format. Crash is available on Dave Anderson's site at the following URL: | |
a7e670d8 | 411 | |
dc851a0f DW |
412 | http://people.redhat.com/~anderson/ |
413 | ||
414 | ||
415 | To Do | |
416 | ===== | |
a7e670d8 | 417 | |
30430134 SH |
418 | 1) Provide relocatable kernels for all architectures to help in maintaining |
419 | multiple kernels for crash_dump, and the same kernel as the system kernel | |
420 | can be used to capture the dump. | |
b089f4a6 | 421 | |
dc851a0f DW |
422 | |
423 | Contact | |
b089f4a6 | 424 | ======= |
dc851a0f | 425 | |
b089f4a6 | 426 | Vivek Goyal (vgoyal@in.ibm.com) |
d58831e4 | 427 | Maneesh Soni (maneesh@in.ibm.com) |
dc851a0f DW |
428 | |
429 | ||
430 | Trademark | |
431 | ========= | |
432 | ||
433 | Linux is a trademark of Linus Torvalds in the United States, other | |
434 | countries, or both. |