kdump.txt 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434
  1. ================================================================
  2. Documentation for Kdump - The kexec-based Crash Dumping Solution
  3. ================================================================
  4. This document includes overview, setup and installation, and analysis
  5. information.
  6. Overview
  7. ========
  8. Kdump uses kexec to quickly boot to a dump-capture kernel whenever a
  9. dump of the system kernel's memory needs to be taken (for example, when
  10. the system panics). The system kernel's memory image is preserved across
  11. the reboot and is accessible to the dump-capture kernel.
  12. You can use common Linux commands, such as cp and scp, to copy the
  13. memory image to a dump file on the local disk, or across the network to
  14. a remote system.
  15. Kdump and kexec are currently supported on the x86, x86_64, ppc64 and ia64
  16. architectures.
  17. When the system kernel boots, it reserves a small section of memory for
  18. the dump-capture kernel. This ensures that ongoing Direct Memory Access
  19. (DMA) from the system kernel does not corrupt the dump-capture kernel.
  20. The kexec -p command loads the dump-capture kernel into this reserved
  21. memory.
  22. On x86 machines, the first 640 KB of physical memory is needed to boot,
  23. regardless of where the kernel loads. Therefore, kexec backs up this
  24. region just before rebooting into the dump-capture kernel.
  25. Similarly on PPC64 machines first 32KB of physical memory is needed for
  26. booting regardless of where the kernel is loaded and to support 64K page
  27. size kexec backs up the first 64KB memory.
  28. All of the necessary information about the system kernel's core image is
  29. encoded in the ELF format, and stored in a reserved area of memory
  30. before a crash. The physical address of the start of the ELF header is
  31. passed to the dump-capture kernel through the elfcorehdr= boot
  32. parameter.
  33. With the dump-capture kernel, you can access the memory image, or "old
  34. memory," in two ways:
  35. - Through a /dev/oldmem device interface. A capture utility can read the
  36. device file and write out the memory in raw format. This is a raw dump
  37. of memory. Analysis and capture tools must be intelligent enough to
  38. determine where to look for the right information.
  39. - Through /proc/vmcore. This exports the dump as an ELF-format file that
  40. you can write out using file copy commands such as cp or scp. Further,
  41. you can use analysis tools such as the GNU Debugger (GDB) and the Crash
  42. tool to debug the dump file. This method ensures that the dump pages are
  43. correctly ordered.
  44. Setup and Installation
  45. ======================
  46. Install kexec-tools
  47. -------------------
  48. 1) Login as the root user.
  49. 2) Download the kexec-tools user-space package from the following URL:
  50. http://www.kernel.org/pub/linux/kernel/people/horms/kexec-tools/kexec-tools-testing.tar.gz
  51. This is a symlink to the latest version, which at the time of writing is
  52. 20061214, the only release of kexec-tools-testing so far. As other versions
  53. are made released, the older onese will remain available at
  54. http://www.kernel.org/pub/linux/kernel/people/horms/kexec-tools/
  55. Note: Latest kexec-tools-testing git tree is available at
  56. git://git.kernel.org/pub/scm/linux/kernel/git/horms/kexec-tools-testing.git
  57. or
  58. http://www.kernel.org/git/?p=linux/kernel/git/horms/kexec-tools-testing.git;a=summary
  59. 3) Unpack the tarball with the tar command, as follows:
  60. tar xvpzf kexec-tools-testing.tar.gz
  61. 4) Change to the kexec-tools directory, as follows:
  62. cd kexec-tools-testing-VERSION
  63. 5) Configure the package, as follows:
  64. ./configure
  65. 6) Compile the package, as follows:
  66. make
  67. 7) Install the package, as follows:
  68. make install
  69. Build the system and dump-capture kernels
  70. -----------------------------------------
  71. There are two possible methods of using Kdump.
  72. 1) Build a separate custom dump-capture kernel for capturing the
  73. kernel core dump.
  74. 2) Or use the system kernel binary itself as dump-capture kernel and there is
  75. no need to build a separate dump-capture kernel. This is possible
  76. only with the architecutres which support a relocatable kernel. As
  77. of today i386 and ia64 architectures support relocatable kernel.
  78. Building a relocatable kernel is advantageous from the point of view that
  79. one does not have to build a second kernel for capturing the dump. But
  80. at the same time one might want to build a custom dump capture kernel
  81. suitable to his needs.
  82. Following are the configuration setting required for system and
  83. dump-capture kernels for enabling kdump support.
  84. System kernel config options
  85. ----------------------------
  86. 1) Enable "kexec system call" in "Processor type and features."
  87. CONFIG_KEXEC=y
  88. 2) Enable "sysfs file system support" in "Filesystem" -> "Pseudo
  89. filesystems." This is usually enabled by default.
  90. CONFIG_SYSFS=y
  91. Note that "sysfs file system support" might not appear in the "Pseudo
  92. filesystems" menu if "Configure standard kernel features (for small
  93. systems)" is not enabled in "General Setup." In this case, check the
  94. .config file itself to ensure that sysfs is turned on, as follows:
  95. grep 'CONFIG_SYSFS' .config
  96. 3) Enable "Compile the kernel with debug info" in "Kernel hacking."
  97. CONFIG_DEBUG_INFO=Y
  98. This causes the kernel to be built with debug symbols. The dump
  99. analysis tools require a vmlinux with debug symbols in order to read
  100. and analyze a dump file.
  101. Dump-capture kernel config options (Arch Independent)
  102. -----------------------------------------------------
  103. 1) Enable "kernel crash dumps" support under "Processor type and
  104. features":
  105. CONFIG_CRASH_DUMP=y
  106. 2) Enable "/proc/vmcore support" under "Filesystems" -> "Pseudo filesystems".
  107. CONFIG_PROC_VMCORE=y
  108. (CONFIG_PROC_VMCORE is set by default when CONFIG_CRASH_DUMP is selected.)
  109. Dump-capture kernel config options (Arch Dependent, i386)
  110. --------------------------------------------------------
  111. 1) On x86, enable high memory support under "Processor type and
  112. features":
  113. CONFIG_HIGHMEM64G=y
  114. or
  115. CONFIG_HIGHMEM4G
  116. 2) On x86 and x86_64, disable symmetric multi-processing support
  117. under "Processor type and features":
  118. CONFIG_SMP=n
  119. (If CONFIG_SMP=y, then specify maxcpus=1 on the kernel command line
  120. when loading the dump-capture kernel, see section "Load the Dump-capture
  121. Kernel".)
  122. 3) If one wants to build and use a relocatable kernel,
  123. Enable "Build a relocatable kernel" support under "Processor type and
  124. features"
  125. CONFIG_RELOCATABLE=y
  126. 4) Use a suitable value for "Physical address where the kernel is
  127. loaded" (under "Processor type and features"). This only appears when
  128. "kernel crash dumps" is enabled. A suitable value depends upon
  129. whether kernel is relocatable or not.
  130. If you are using a relocatable kernel use CONFIG_PHYSICAL_START=0x100000
  131. This will compile the kernel for physical address 1MB, but given the fact
  132. kernel is relocatable, it can be run from any physical address hence
  133. kexec boot loader will load it in memory region reserved for dump-capture
  134. kernel.
  135. Otherwise it should be the start of memory region reserved for
  136. second kernel using boot parameter "crashkernel=Y@X". Here X is
  137. start of memory region reserved for dump-capture kernel.
  138. Generally X is 16MB (0x1000000). So you can set
  139. CONFIG_PHYSICAL_START=0x1000000
  140. 5) Make and install the kernel and its modules. DO NOT add this kernel
  141. to the boot loader configuration files.
  142. Dump-capture kernel config options (Arch Dependent, x86_64)
  143. ----------------------------------------------------------
  144. 1) On x86 and x86_64, disable symmetric multi-processing support
  145. under "Processor type and features":
  146. CONFIG_SMP=n
  147. (If CONFIG_SMP=y, then specify maxcpus=1 on the kernel command line
  148. when loading the dump-capture kernel, see section "Load the Dump-capture
  149. Kernel".)
  150. 2) Use a suitable value for "Physical address where the kernel is
  151. loaded" (under "Processor type and features"). This only appears when
  152. "kernel crash dumps" is enabled. By default this value is 0x1000000
  153. (16MB). It should be the same as X in the "crashkernel=Y@X" boot
  154. parameter.
  155. For x86_64, normally "CONFIG_PHYSICAL_START=0x1000000".
  156. 3) Make and install the kernel and its modules. DO NOT add this kernel
  157. to the boot loader configuration files.
  158. Dump-capture kernel config options (Arch Dependent, ppc64)
  159. ----------------------------------------------------------
  160. * Make and install the kernel and its modules. DO NOT add this kernel
  161. to the boot loader configuration files.
  162. Dump-capture kernel config options (Arch Dependent, ia64)
  163. ----------------------------------------------------------
  164. - No specific options are required to create a dump-capture kernel
  165. for ia64, other than those specified in the arch idependent section
  166. above. This means that it is possible to use the system kernel
  167. as a dump-capture kernel if desired.
  168. The crashkernel region can be automatically placed by the system
  169. kernel at run time. This is done by specifying the base address as 0,
  170. or omitting it all together.
  171. crashkernel=256M@0
  172. or
  173. crashkernel=256M
  174. If the start address is specified, note that the start address of the
  175. kernel will be aligned to 64Mb, so if the start address is not then
  176. any space below the alignment point will be wasted.
  177. Boot into System Kernel
  178. =======================
  179. 1) Update the boot loader (such as grub, yaboot, or lilo) configuration
  180. files as necessary.
  181. 2) Boot the system kernel with the boot parameter "crashkernel=Y@X",
  182. where Y specifies how much memory to reserve for the dump-capture kernel
  183. and X specifies the beginning of this reserved memory. For example,
  184. "crashkernel=64M@16M" tells the system kernel to reserve 64 MB of memory
  185. starting at physical address 0x01000000 (16MB) for the dump-capture kernel.
  186. On x86 and x86_64, use "crashkernel=64M@16M".
  187. On ppc64, use "crashkernel=128M@32M".
  188. On ia64, 256M@256M is a generous value that typically works.
  189. The region may be automatically placed on ia64, see the
  190. dump-capture kernel config option notes above.
  191. Load the Dump-capture Kernel
  192. ============================
  193. After booting to the system kernel, dump-capture kernel needs to be
  194. loaded.
  195. Based on the architecture and type of image (relocatable or not), one
  196. can choose to load the uncompressed vmlinux or compressed bzImage/vmlinuz
  197. of dump-capture kernel. Following is the summary.
  198. For i386:
  199. - Use vmlinux if kernel is not relocatable.
  200. - Use bzImage/vmlinuz if kernel is relocatable.
  201. For x86_64:
  202. - Use vmlinux
  203. For ppc64:
  204. - Use vmlinux
  205. For ia64:
  206. - Use vmlinux or vmlinuz.gz
  207. If you are using a uncompressed vmlinux image then use following command
  208. to load dump-capture kernel.
  209. kexec -p <dump-capture-kernel-vmlinux-image> \
  210. --initrd=<initrd-for-dump-capture-kernel> --args-linux \
  211. --append="root=<root-dev> <arch-specific-options>"
  212. If you are using a compressed bzImage/vmlinuz, then use following command
  213. to load dump-capture kernel.
  214. kexec -p <dump-capture-kernel-bzImage> \
  215. --initrd=<initrd-for-dump-capture-kernel> \
  216. --append="root=<root-dev> <arch-specific-options>"
  217. Please note, that --args-linux does not need to be specified for ia64.
  218. It is planned to make this a no-op on that architecture, but for now
  219. it should be omitted
  220. Following are the arch specific command line options to be used while
  221. loading dump-capture kernel.
  222. For i386, x86_64 and ia64:
  223. "1 irqpoll maxcpus=1"
  224. For ppc64:
  225. "1 maxcpus=1 noirqdistrib"
  226. Notes on loading the dump-capture kernel:
  227. * By default, the ELF headers are stored in ELF64 format to support
  228. systems with more than 4GB memory. The --elf32-core-headers option can
  229. be used to force the generation of ELF32 headers. This is necessary
  230. because GDB currently cannot open vmcore files with ELF64 headers on
  231. 32-bit systems. ELF32 headers can be used on non-PAE systems (that is,
  232. less than 4GB of memory).
  233. * The "irqpoll" boot parameter reduces driver initialization failures
  234. due to shared interrupts in the dump-capture kernel.
  235. * You must specify <root-dev> in the format corresponding to the root
  236. device name in the output of mount command.
  237. * Boot parameter "1" boots the dump-capture kernel into single-user
  238. mode without networking. If you want networking, use "3".
  239. * We generally don' have to bring up a SMP kernel just to capture the
  240. dump. Hence generally it is useful either to build a UP dump-capture
  241. kernel or specify maxcpus=1 option while loading dump-capture kernel.
  242. Kernel Panic
  243. ============
  244. After successfully loading the dump-capture kernel as previously
  245. described, the system will reboot into the dump-capture kernel if a
  246. system crash is triggered. Trigger points are located in panic(),
  247. die(), die_nmi() and in the sysrq handler (ALT-SysRq-c).
  248. The following conditions will execute a crash trigger point:
  249. If a hard lockup is detected and "NMI watchdog" is configured, the system
  250. will boot into the dump-capture kernel ( die_nmi() ).
  251. If die() is called, and it happens to be a thread with pid 0 or 1, or die()
  252. is called inside interrupt context or die() is called and panic_on_oops is set,
  253. the system will boot into the dump-capture kernel.
  254. On powererpc systems when a soft-reset is generated, die() is called by all cpus
  255. and the system will boot into the dump-capture kernel.
  256. For testing purposes, you can trigger a crash by using "ALT-SysRq-c",
  257. "echo c > /proc/sysrq-trigger" or write a module to force the panic.
  258. Write Out the Dump File
  259. =======================
  260. After the dump-capture kernel is booted, write out the dump file with
  261. the following command:
  262. cp /proc/vmcore <dump-file>
  263. You can also access dumped memory as a /dev/oldmem device for a linear
  264. and raw view. To create the device, use the following command:
  265. mknod /dev/oldmem c 1 12
  266. Use the dd command with suitable options for count, bs, and skip to
  267. access specific portions of the dump.
  268. To see the entire memory, use the following command:
  269. dd if=/dev/oldmem of=oldmem.001
  270. Analysis
  271. ========
  272. Before analyzing the dump image, you should reboot into a stable kernel.
  273. You can do limited analysis using GDB on the dump file copied out of
  274. /proc/vmcore. Use the debug vmlinux built with -g and run the following
  275. command:
  276. gdb vmlinux <dump-file>
  277. Stack trace for the task on processor 0, register display, and memory
  278. display work fine.
  279. Note: GDB cannot analyze core files generated in ELF64 format for x86.
  280. On systems with a maximum of 4GB of memory, you can generate
  281. ELF32-format headers using the --elf32-core-headers kernel option on the
  282. dump kernel.
  283. You can also use the Crash utility to analyze dump files in Kdump
  284. format. Crash is available on Dave Anderson's site at the following URL:
  285. http://people.redhat.com/~anderson/
  286. To Do
  287. =====
  288. 1) Provide relocatable kernels for all architectures to help in maintaining
  289. multiple kernels for crash_dump, and the same kernel as the system kernel
  290. can be used to capture the dump.
  291. Contact
  292. =======
  293. Vivek Goyal (vgoyal@in.ibm.com)
  294. Maneesh Soni (maneesh@in.ibm.com)
  295. Trademark
  296. =========
  297. Linux is a trademark of Linus Torvalds in the United States, other
  298. countries, or both.