linux/Documentation/kdump/kdump.txt
<<
>>
Prefs
   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 and ia64
  21architectures.
  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
  37All 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.
  42
  43With the dump-capture kernel, you can access the memory image, or "old
  44memory," 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
  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://www.kernel.org/pub/linux/kernel/people/horms/kexec-tools/kexec-tools.tar.gz
  69
  70This is a symlink to the latest version.
  71
  72The latest kexec-tools git tree is available at:
  73
  74git://git.kernel.org/pub/scm/linux/kernel/git/horms/kexec-tools.git
  75or
  76http://www.kernel.org/git/?p=linux/kernel/git/horms/kexec-tools.git
  77
  78More information about kexec-tools can be found at
  79http://www.kernel.org/pub/linux/kernel/people/horms/kexec-tools/README.html
  80
  813) Unpack the tarball with the tar command, as follows:
  82
  83   tar xvpzf kexec-tools.tar.gz
  84
  854) Change to the kexec-tools directory, as follows:
  86
  87   cd kexec-tools-VERSION
  88
  895) Configure the package, as follows:
  90
  91   ./configure
  92
  936) Compile the package, as follows:
  94
  95   make
  96
  977) Install the package, as follows:
  98
  99   make install
 100
 101
 102Build the system and dump-capture kernels
 103-----------------------------------------
 104There are two possible methods of using Kdump.
 105
 1061) Build a separate custom dump-capture kernel for capturing the
 107   kernel core dump.
 108
 1092) 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 architectures which support a relocatable kernel. As
 112   of today, i386, x86_64, ppc64 and ia64 architectures support relocatable
 113   kernel.
 114
 115Building a relocatable kernel is advantageous from the point of view that
 116one does not have to build a second kernel for capturing the dump. But
 117at the same time one might want to build a custom dump capture kernel
 118suitable to his needs.
 119
 120Following are the configuration setting required for system and
 121dump-capture kernels for enabling kdump support.
 122
 123System kernel config options
 124----------------------------
 125
 1261) Enable "kexec system call" in "Processor type and features."
 127
 128   CONFIG_KEXEC=y
 129
 1302) Enable "sysfs file system support" in "Filesystem" -> "Pseudo
 131   filesystems." This is usually enabled by default.
 132
 133   CONFIG_SYSFS=y
 134
 135   Note that "sysfs file system support" might not appear in the "Pseudo
 136   filesystems" menu if "Configure standard kernel features (for small
 137   systems)" is not enabled in "General Setup." In this case, check the
 138   .config file itself to ensure that sysfs is turned on, as follows:
 139
 140   grep 'CONFIG_SYSFS' .config
 141
 1423) Enable "Compile the kernel with debug info" in "Kernel hacking."
 143
 144   CONFIG_DEBUG_INFO=Y
 145
 146   This causes the kernel to be built with debug symbols. The dump
 147   analysis tools require a vmlinux with debug symbols in order to read
 148   and analyze a dump file.
 149
 150Dump-capture kernel config options (Arch Independent)
 151-----------------------------------------------------
 152
 1531) Enable "kernel crash dumps" support under "Processor type and
 154   features":
 155
 156   CONFIG_CRASH_DUMP=y
 157
 1582) Enable "/proc/vmcore support" under "Filesystems" -> "Pseudo filesystems".
 159
 160   CONFIG_PROC_VMCORE=y
 161   (CONFIG_PROC_VMCORE is set by default when CONFIG_CRASH_DUMP is selected.)
 162
 163Dump-capture kernel config options (Arch Dependent, i386 and x86_64)
 164--------------------------------------------------------------------
 165
 1661) On i386, enable high memory support under "Processor type and
 167   features":
 168
 169   CONFIG_HIGHMEM64G=y
 170   or
 171   CONFIG_HIGHMEM4G
 172
 1732) On i386 and x86_64, disable symmetric multi-processing support
 174   under "Processor type and features":
 175
 176   CONFIG_SMP=n
 177
 178   (If CONFIG_SMP=y, then specify maxcpus=1 on the kernel command line
 179   when loading the dump-capture kernel, see section "Load the Dump-capture
 180   Kernel".)
 181
 1823) If one wants to build and use a relocatable kernel,
 183   Enable "Build a relocatable kernel" support under "Processor type and
 184   features"
 185
 186   CONFIG_RELOCATABLE=y
 187
 1884) Use a suitable value for "Physical address where the kernel is
 189   loaded" (under "Processor type and features"). This only appears when
 190   "kernel crash dumps" is enabled. A suitable value depends upon
 191   whether kernel is relocatable or not.
 192
 193   If you are using a relocatable kernel use CONFIG_PHYSICAL_START=0x100000
 194   This will compile the kernel for physical address 1MB, but given the fact
 195   kernel is relocatable, it can be run from any physical address hence
 196   kexec boot loader will load it in memory region reserved for dump-capture
 197   kernel.
 198
 199   Otherwise it should be the start of memory region reserved for
 200   second kernel using boot parameter "crashkernel=Y@X". Here X is
 201   start of memory region reserved for dump-capture kernel.
 202   Generally X is 16MB (0x1000000). So you can set
 203   CONFIG_PHYSICAL_START=0x1000000
 204
 2055) Make and install the kernel and its modules. DO NOT add this kernel
 206   to the boot loader configuration files.
 207
 208Dump-capture kernel config options (Arch Dependent, ppc64)
 209----------------------------------------------------------
 210
 2111) Enable "Build a kdump crash kernel" support under "Kernel" options:
 212
 213   CONFIG_CRASH_DUMP=y
 214
 2152)   Enable "Build a relocatable kernel" support
 216
 217   CONFIG_RELOCATABLE=y
 218
 219   Make and install the kernel and its modules.
 220
 221Dump-capture kernel config options (Arch Dependent, ia64)
 222----------------------------------------------------------
 223
 224- No specific options are required to create a dump-capture kernel
 225  for ia64, other than those specified in the arch independent section
 226  above. This means that it is possible to use the system kernel
 227  as a dump-capture kernel if desired.
 228
 229  The crashkernel region can be automatically placed by the system
 230  kernel at run time. This is done by specifying the base address as 0,
 231  or omitting it all together.
 232
 233  crashkernel=256M@0
 234  or
 235  crashkernel=256M
 236
 237  If the start address is specified, note that the start address of the
 238  kernel will be aligned to 64Mb, so if the start address is not then
 239  any space below the alignment point will be wasted.
 240
 241
 242Extended crashkernel syntax
 243===========================
 244
 245While the "crashkernel=size[@offset]" syntax is sufficient for most
 246configurations, sometimes it's handy to have the reserved memory dependent
 247on the value of System RAM -- that's mostly for distributors that pre-setup
 248the kernel command line to avoid a unbootable system after some memory has
 249been removed from the machine.
 250
 251The syntax is:
 252
 253    crashkernel=<range1>:<size1>[,<range2>:<size2>,...][@offset]
 254    range=start-[end]
 255
 256    'start' is inclusive and 'end' is exclusive.
 257
 258For example:
 259
 260    crashkernel=512M-2G:64M,2G-:128M
 261
 262This would mean:
 263
 264    1) if the RAM is smaller than 512M, then don't reserve anything
 265       (this is the "rescue" case)
 266    2) if the RAM size is between 512M and 2G (exclusive), then reserve 64M
 267    3) if the RAM size is larger than 2G, then reserve 128M
 268
 269
 270
 271Boot into System Kernel
 272=======================
 273
 2741) Update the boot loader (such as grub, yaboot, or lilo) configuration
 275   files as necessary.
 276
 2772) Boot the system kernel with the boot parameter "crashkernel=Y@X",
 278   where Y specifies how much memory to reserve for the dump-capture kernel
 279   and X specifies the beginning of this reserved memory. For example,
 280   "crashkernel=64M@16M" tells the system kernel to reserve 64 MB of memory
 281   starting at physical address 0x01000000 (16MB) for the dump-capture kernel.
 282
 283   On x86 and x86_64, use "crashkernel=64M@16M".
 284
 285   On ppc64, use "crashkernel=128M@32M".
 286
 287   On ia64, 256M@256M is a generous value that typically works.
 288   The region may be automatically placed on ia64, see the
 289   dump-capture kernel config option notes above.
 290
 291Load the Dump-capture Kernel
 292============================
 293
 294After booting to the system kernel, dump-capture kernel needs to be
 295loaded.
 296
 297Based on the architecture and type of image (relocatable or not), one
 298can choose to load the uncompressed vmlinux or compressed bzImage/vmlinuz
 299of dump-capture kernel. Following is the summary.
 300
 301For i386 and x86_64:
 302        - Use vmlinux if kernel is not relocatable.
 303        - Use bzImage/vmlinuz if kernel is relocatable.
 304For ppc64:
 305        - Use vmlinux
 306For ia64:
 307        - Use vmlinux or vmlinuz.gz
 308
 309
 310If you are using a uncompressed vmlinux image then use following command
 311to load dump-capture kernel.
 312
 313   kexec -p <dump-capture-kernel-vmlinux-image> \
 314   --initrd=<initrd-for-dump-capture-kernel> --args-linux \
 315   --append="root=<root-dev> <arch-specific-options>"
 316
 317If you are using a compressed bzImage/vmlinuz, then use following command
 318to load dump-capture kernel.
 319
 320   kexec -p <dump-capture-kernel-bzImage> \
 321   --initrd=<initrd-for-dump-capture-kernel> \
 322   --append="root=<root-dev> <arch-specific-options>"
 323
 324Please note, that --args-linux does not need to be specified for ia64.
 325It is planned to make this a no-op on that architecture, but for now
 326it should be omitted
 327
 328Following are the arch specific command line options to be used while
 329loading dump-capture kernel.
 330
 331For i386, x86_64 and ia64:
 332        "1 irqpoll maxcpus=1 reset_devices"
 333
 334For ppc64:
 335        "1 maxcpus=1 noirqdistrib reset_devices"
 336
 337
 338Notes on loading the dump-capture kernel:
 339
 340* By default, the ELF headers are stored in ELF64 format to support
 341  systems with more than 4GB memory. On i386, kexec automatically checks if
 342  the physical RAM size exceeds the 4 GB limit and if not, uses ELF32.
 343  So, on non-PAE systems, ELF32 is always used.
 344
 345  The --elf32-core-headers option can be used to force the generation of ELF32
 346  headers. This is necessary because GDB currently cannot open vmcore files
 347  with ELF64 headers on 32-bit systems.
 348
 349* The "irqpoll" boot parameter reduces driver initialization failures
 350  due to shared interrupts in the dump-capture kernel.
 351
 352* You must specify <root-dev> in the format corresponding to the root
 353  device name in the output of mount command.
 354
 355* Boot parameter "1" boots the dump-capture kernel into single-user
 356  mode without networking. If you want networking, use "3".
 357
 358* We generally don' have to bring up a SMP kernel just to capture the
 359  dump. Hence generally it is useful either to build a UP dump-capture
 360  kernel or specify maxcpus=1 option while loading dump-capture kernel.
 361
 362Kernel Panic
 363============
 364
 365After successfully loading the dump-capture kernel as previously
 366described, the system will reboot into the dump-capture kernel if a
 367system crash is triggered.  Trigger points are located in panic(),
 368die(), die_nmi() and in the sysrq handler (ALT-SysRq-c).
 369
 370The following conditions will execute a crash trigger point:
 371
 372If a hard lockup is detected and "NMI watchdog" is configured, the system
 373will boot into the dump-capture kernel ( die_nmi() ).
 374
 375If die() is called, and it happens to be a thread with pid 0 or 1, or die()
 376is called inside interrupt context or die() is called and panic_on_oops is set,
 377the system will boot into the dump-capture kernel.
 378
 379On powerpc systems when a soft-reset is generated, die() is called by all cpus
 380and the system will boot into the dump-capture kernel.
 381
 382For testing purposes, you can trigger a crash by using "ALT-SysRq-c",
 383"echo c > /proc/sysrq-trigger" or write a module to force the panic.
 384
 385Write Out the Dump File
 386=======================
 387
 388After the dump-capture kernel is booted, write out the dump file with
 389the following command:
 390
 391   cp /proc/vmcore <dump-file>
 392
 393You can also access dumped memory as a /dev/oldmem device for a linear
 394and raw view. To create the device, use the following command:
 395
 396    mknod /dev/oldmem c 1 12
 397
 398Use the dd command with suitable options for count, bs, and skip to
 399access specific portions of the dump.
 400
 401To see the entire memory, use the following command:
 402
 403   dd if=/dev/oldmem of=oldmem.001
 404
 405
 406Analysis
 407========
 408
 409Before analyzing the dump image, you should reboot into a stable kernel.
 410
 411You can do limited analysis using GDB on the dump file copied out of
 412/proc/vmcore. Use the debug vmlinux built with -g and run the following
 413command:
 414
 415   gdb vmlinux <dump-file>
 416
 417Stack trace for the task on processor 0, register display, and memory
 418display work fine.
 419
 420Note: GDB cannot analyze core files generated in ELF64 format for x86.
 421On systems with a maximum of 4GB of memory, you can generate
 422ELF32-format headers using the --elf32-core-headers kernel option on the
 423dump kernel.
 424
 425You can also use the Crash utility to analyze dump files in Kdump
 426format. Crash is available on Dave Anderson's site at the following URL:
 427
 428   http://people.redhat.com/~anderson/
 429
 430
 431To Do
 432=====
 433
 4341) Provide relocatable kernels for all architectures to help in maintaining
 435   multiple kernels for crash_dump, and the same kernel as the system kernel
 436   can be used to capture the dump.
 437
 438
 439Contact
 440=======
 441
 442Vivek Goyal (vgoyal@in.ibm.com)
 443Maneesh Soni (maneesh@in.ibm.com)
 444
 445
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.