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 and arm 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 74git://git.kernel.org/pub/scm/utils/kernel/kexec/kexec-tools.git 75and 76http://www.kernel.org/pub/scm/utils/kernel/kexec/kexec-tools.git 77 78There is also a gitweb interface available at 79http://www.kernel.org/git/?p=utils/kernel/kexec/kexec-tools.git 80 81More information about kexec-tools can be found at 82http://horms.net/projects/kexec/ 83 843) Unpack the tarball with the tar command, as follows: 85 86 tar xvpzf kexec-tools.tar.gz 87 884) Change to the kexec-tools directory, as follows: 89 90 cd kexec-tools-VERSION 91 925) Configure the package, as follows: 93 94 ./configure 95 966) Compile the package, as follows: 97 98 make 99 1007) Install the package, as follows: 101 102 make install 103 104 105Build the system and dump-capture kernels 106----------------------------------------- 107There are two possible methods of using Kdump. 108 1091) Build a separate custom dump-capture kernel for capturing the 110 kernel core dump. 111 1122) Or use the system kernel binary itself as dump-capture kernel and there is 113 no need to build a separate dump-capture kernel. This is possible 114 only with the architectures which support a relocatable kernel. As 115 of today, i386, x86_64, ppc64, ia64 and arm architectures support relocatable 116 kernel. 117 118Building a relocatable kernel is advantageous from the point of view that 119one does not have to build a second kernel for capturing the dump. But 120at the same time one might want to build a custom dump capture kernel 121suitable to his needs. 122 123Following are the configuration setting required for system and 124dump-capture kernels for enabling kdump support. 125 126System kernel config options 127---------------------------- 128 1291) Enable "kexec system call" in "Processor type and features." 130 131 CONFIG_KEXEC=y 132 1332) Enable "sysfs file system support" in "Filesystem" -> "Pseudo 134 filesystems." This is usually enabled by default. 135 136 CONFIG_SYSFS=y 137 138 Note that "sysfs file system support" might not appear in the "Pseudo 139 filesystems" menu if "Configure standard kernel features (for small 140 systems)" is not enabled in "General Setup." In this case, check the 141 .config file itself to ensure that sysfs is turned on, as follows: 142 143 grep 'CONFIG_SYSFS' .config 144 1453) Enable "Compile the kernel with debug info" in "Kernel hacking." 146 147 CONFIG_DEBUG_INFO=Y 148 149 This causes the kernel to be built with debug symbols. The dump 150 analysis tools require a vmlinux with debug symbols in order to read 151 and analyze a dump file. 152 153Dump-capture kernel config options (Arch Independent) 154----------------------------------------------------- 155 1561) Enable "kernel crash dumps" support under "Processor type and 157 features": 158 159 CONFIG_CRASH_DUMP=y 160 1612) Enable "/proc/vmcore support" under "Filesystems" -> "Pseudo filesystems". 162 163 CONFIG_PROC_VMCORE=y 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 or 174 CONFIG_HIGHMEM4G 175 1762) On i386 and x86_64, disable symmetric multi-processing support 177 under "Processor type and features": 178 179 CONFIG_SMP=n 180 181 (If CONFIG_SMP=y, then specify maxcpus=1 on the kernel command line 182 when loading the dump-capture kernel, see section "Load the Dump-capture 183 Kernel".) 184 1853) If one wants to build and use a relocatable kernel, 186 Enable "Build a relocatable kernel" support under "Processor type and 187 features" 188 189 CONFIG_RELOCATABLE=y 190 1914) Use a suitable value for "Physical address where the kernel is 192 loaded" (under "Processor type and features"). This only appears when 193 "kernel crash dumps" is enabled. A suitable value depends upon 194 whether kernel is relocatable or not. 195 196 If you are using a relocatable kernel use CONFIG_PHYSICAL_START=0x100000 197 This will compile the kernel for physical address 1MB, but given the fact 198 kernel is relocatable, it can be run from any physical address hence 199 kexec boot loader will load it in memory region reserved for dump-capture 200 kernel. 201 202 Otherwise it should be the start of memory region reserved for 203 second kernel using boot parameter "crashkernel=Y@X". Here X is 204 start of memory region reserved for dump-capture kernel. 205 Generally X is 16MB (0x1000000). So you can set 206 CONFIG_PHYSICAL_START=0x1000000 207 2085) Make and install the kernel and its modules. DO NOT add this kernel 209 to the boot loader configuration files. 210 211Dump-capture kernel config options (Arch Dependent, ppc64) 212---------------------------------------------------------- 213 2141) Enable "Build a kdump crash kernel" support under "Kernel" options: 215 216 CONFIG_CRASH_DUMP=y 217 2182) Enable "Build a relocatable kernel" support 219 220 CONFIG_RELOCATABLE=y 221 222 Make and install the kernel and its modules. 223 224Dump-capture kernel config options (Arch Dependent, ia64) 225---------------------------------------------------------- 226 227- No specific options are required to create a dump-capture kernel 228 for ia64, other than those specified in the arch independent section 229 above. This means that it is possible to use the system kernel 230 as a dump-capture kernel if desired. 231 232 The crashkernel region can be automatically placed by the system 233 kernel at run time. This is done by specifying the base address as 0, 234 or omitting it all together. 235 236 crashkernel=256M@0 237 or 238 crashkernel=256M 239 240 If the start address is specified, note that the start address of the 241 kernel will be aligned to 64Mb, so if the start address is not then 242 any space below the alignment point will be wasted. 243 244Dump-capture kernel config options (Arch Dependent, arm) 245---------------------------------------------------------- 246 247- To use a relocatable kernel, 248 Enable "AUTO_ZRELADDR" support under "Boot" options: 249 250 AUTO_ZRELADDR=y 251 252Extended crashkernel syntax 253=========================== 254 255While the "crashkernel=size[@offset]" syntax is sufficient for most 256configurations, sometimes it's handy to have the reserved memory dependent 257on the value of System RAM -- that's mostly for distributors that pre-setup 258the kernel command line to avoid a unbootable system after some memory has 259been removed from the machine. 260 261The syntax is: 262 263 crashkernel=<range1>:<size1>[,<range2>:<size2>,...][@offset] 264 range=start-[end] 265 266For example: 267 268 crashkernel=512M-2G:64M,2G-:128M 269 270This would mean: 271 272 1) if the RAM is smaller than 512M, then don't reserve anything 273 (this is the "rescue" case) 274 2) if the RAM size is between 512M and 2G (exclusive), then reserve 64M 275 3) if the RAM size is larger than 2G, then reserve 128M 276 277 278 279Boot into System Kernel 280======================= 281 2821) Update the boot loader (such as grub, yaboot, or lilo) configuration 283 files as necessary. 284 2852) Boot the system kernel with the boot parameter "crashkernel=Y@X", 286 where Y specifies how much memory to reserve for the dump-capture kernel 287 and X specifies the beginning of this reserved memory. For example, 288 "crashkernel=64M@16M" tells the system kernel to reserve 64 MB of memory 289 starting at physical address 0x01000000 (16MB) for the dump-capture kernel. 290 291 On x86 and x86_64, use "crashkernel=64M@16M". 292 293 On ppc64, use "crashkernel=128M@32M". 294 295 On ia64, 256M@256M is a generous value that typically works. 296 The region may be automatically placed on ia64, see the 297 dump-capture kernel config option notes above. 298 If use sparse memory, the size should be rounded to GRANULE boundaries. 299 300 On s390x, typically use "crashkernel=xxM". The value of xx is dependent 301 on the memory consumption of the kdump system. In general this is not 302 dependent on the memory size of the production system. 303 304 On arm, the use of "crashkernel=Y@X" is no longer necessary; the 305 kernel will automatically locate the crash kernel image within the 306 first 512MB of RAM if X is not given. 307 308 309Load the Dump-capture Kernel 310============================ 311 312After booting to the system kernel, dump-capture kernel needs to be 313loaded. 314 315Based on the architecture and type of image (relocatable or not), one 316can choose to load the uncompressed vmlinux or compressed bzImage/vmlinuz 317of dump-capture kernel. Following is the summary. 318 319For i386 and x86_64: 320 - Use vmlinux if kernel is not relocatable. 321 - Use bzImage/vmlinuz if kernel is relocatable. 322For ppc64: 323 - Use vmlinux 324For ia64: 325 - Use vmlinux or vmlinuz.gz 326For s390x: 327 - Use image or bzImage 328For arm: 329 - Use zImage 330 331If you are using a uncompressed vmlinux image then use following command 332to load dump-capture kernel. 333 334 kexec -p <dump-capture-kernel-vmlinux-image> \ 335 --initrd=<initrd-for-dump-capture-kernel> --args-linux \ 336 --append="root=<root-dev> <arch-specific-options>" 337 338If you are using a compressed bzImage/vmlinuz, then use following command 339to load dump-capture kernel. 340 341 kexec -p <dump-capture-kernel-bzImage> \ 342 --initrd=<initrd-for-dump-capture-kernel> \ 343 --append="root=<root-dev> <arch-specific-options>" 344 345If you are using a compressed zImage, then use following command 346to load dump-capture kernel. 347 348 kexec --type zImage -p <dump-capture-kernel-bzImage> \ 349 --initrd=<initrd-for-dump-capture-kernel> \ 350 --dtb=<dtb-for-dump-capture-kernel> \ 351 --append="root=<root-dev> <arch-specific-options>" 352 353 354Please note, that --args-linux does not need to be specified for ia64. 355It is planned to make this a no-op on that architecture, but for now 356it should be omitted 357 358Following are the arch specific command line options to be used while 359loading dump-capture kernel. 360 361For i386, x86_64 and ia64: 362 "1 irqpoll maxcpus=1 reset_devices" 363 364For ppc64: 365 "1 maxcpus=1 noirqdistrib reset_devices" 366 367For s390x: 368 "1 maxcpus=1 cgroup_disable=memory" 369 370For arm: 371 "1 maxcpus=1 reset_devices" 372 373Notes on loading the dump-capture kernel: 374 375* By default, the ELF headers are stored in ELF64 format to support 376 systems with more than 4GB memory. On i386, kexec automatically checks if 377 the physical RAM size exceeds the 4 GB limit and if not, uses ELF32. 378 So, on non-PAE systems, ELF32 is always used. 379 380 The --elf32-core-headers option can be used to force the generation of ELF32 381 headers. This is necessary because GDB currently cannot open vmcore files 382 with ELF64 headers on 32-bit systems. 383 384* The "irqpoll" boot parameter reduces driver initialization failures 385 due to shared interrupts in the dump-capture kernel. 386 387* You must specify <root-dev> in the format corresponding to the root 388 device name in the output of mount command. 389 390* Boot parameter "1" boots the dump-capture kernel into single-user 391 mode without networking. If you want networking, use "3". 392 393* We generally don' have to bring up a SMP kernel just to capture the 394 dump. Hence generally it is useful either to build a UP dump-capture 395 kernel or specify maxcpus=1 option while loading dump-capture kernel. 396 Note, though maxcpus always works, you had better replace it with 397 nr_cpus to save memory if supported by the current ARCH, such as x86. 398 399* You should enable multi-cpu support in dump-capture kernel if you intend 400 to use multi-thread programs with it, such as parallel dump feature of 401 makedumpfile. Otherwise, the multi-thread program may have a great 402 performance degradation. To enable multi-cpu support, you should bring up an 403 SMP dump-capture kernel and specify maxcpus/nr_cpus, disable_cpu_apicid=[X] 404 options while loading it. 405 406* For s390x there are two kdump modes: If a ELF header is specified with 407 the elfcorehdr= kernel parameter, it is used by the kdump kernel as it 408 is done on all other architectures. If no elfcorehdr= kernel parameter is 409 specified, the s390x kdump kernel dynamically creates the header. The 410 second mode has the advantage that for CPU and memory hotplug, kdump has 411 not to be reloaded with kexec_load(). 412 413* For s390x systems with many attached devices the "cio_ignore" kernel 414 parameter should be used for the kdump kernel in order to prevent allocation 415 of kernel memory for devices that are not relevant for kdump. The same 416 applies to systems that use SCSI/FCP devices. In that case the 417 "allow_lun_scan" zfcp module parameter should be set to zero before 418 setting FCP devices online. 419 420Kernel Panic 421============ 422 423After successfully loading the dump-capture kernel as previously 424described, the system will reboot into the dump-capture kernel if a 425system crash is triggered. Trigger points are located in panic(), 426die(), die_nmi() and in the sysrq handler (ALT-SysRq-c). 427 428The following conditions will execute a crash trigger point: 429 430If a hard lockup is detected and "NMI watchdog" is configured, the system 431will boot into the dump-capture kernel ( die_nmi() ). 432 433If die() is called, and it happens to be a thread with pid 0 or 1, or die() 434is called inside interrupt context or die() is called and panic_on_oops is set, 435the system will boot into the dump-capture kernel. 436 437On powerpc systems when a soft-reset is generated, die() is called by all cpus 438and the system will boot into the dump-capture kernel. 439 440For testing purposes, you can trigger a crash by using "ALT-SysRq-c", 441"echo c > /proc/sysrq-trigger" or write a module to force the panic. 442 443Write Out the Dump File 444======================= 445 446After the dump-capture kernel is booted, write out the dump file with 447the following command: 448 449 cp /proc/vmcore <dump-file> 450 451 452Analysis 453======== 454 455Before analyzing the dump image, you should reboot into a stable kernel. 456 457You can do limited analysis using GDB on the dump file copied out of 458/proc/vmcore. Use the debug vmlinux built with -g and run the following 459command: 460 461 gdb vmlinux <dump-file> 462 463Stack trace for the task on processor 0, register display, and memory 464display work fine. 465 466Note: GDB cannot analyze core files generated in ELF64 format for x86. 467On systems with a maximum of 4GB of memory, you can generate 468ELF32-format headers using the --elf32-core-headers kernel option on the 469dump kernel. 470 471You can also use the Crash utility to analyze dump files in Kdump 472format. Crash is available on Dave Anderson's site at the following URL: 473 474 http://people.redhat.com/~anderson/ 475 476Trigger Kdump on WARN() 477======================= 478 479The kernel parameter, panic_on_warn, calls panic() in all WARN() paths. This 480will cause a kdump to occur at the panic() call. In cases where a user wants 481to specify this during runtime, /proc/sys/kernel/panic_on_warn can be set to 1 482to achieve the same behaviour. 483 484Contact 485======= 486 487Vivek Goyal (vgoyal@redhat.com) 488Maneesh Soni (maneesh@in.ibm.com) 489 490