Commit 8bb31b9d5340ed3dfef45d322f59fcf18a0d598b

Authored by Ankita Garg
Committed by Linus Torvalds
1 parent 99219a3fbc

[PATCH] Linux Kernel Dump Test Module

A simple module to test Linux Kernel Dump mechanism.  This module uses
jprobes to install/activate pre-defined crash points.  At different crash
points, various types of crashing scenarios are created like a BUG(),
panic(), exception, recursive loop and stack overflow.  The user can
activate a crash point with specific type by providing parameters at the
time of module insertion.  Please see the file header for usage
information.  The module is based on the Linux Kernel Dump Test Tool by
Fernando <http://lkdtt.sourceforge.net>.

This module could be merged with mainline. Jprobes is used here so that the
context in which crash point is hit, could be maintained. This implements
all the crash points as done by LKDTT except the one in the middle of
tasklet_action().

Signed-off-by: Ankita Garg <ankita@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Showing 3 changed files with 358 additions and 1 deletions Side-by-side Diff

drivers/misc/Makefile
... ... @@ -3,6 +3,7 @@
3 3 #
4 4 obj- := misc.o # Dummy rule to force built-in.o to be made
5 5  
6   -obj-$(CONFIG_IBM_ASM) += ibmasm/
  6 +obj-$(CONFIG_IBM_ASM) += ibmasm/
7 7 obj-$(CONFIG_HDPU_FEATURES) += hdpuftrs/
  8 +obj-$(CONFIG_LKDTM) += lkdtm.o
drivers/misc/lkdtm.c
  1 +/*
  2 + * Kprobe module for testing crash dumps
  3 + *
  4 + * This program is free software; you can redistribute it and/or modify
  5 + * it under the terms of the GNU General Public License as published by
  6 + * the Free Software Foundation; either version 2 of the License, or
  7 + * (at your option) any later version.
  8 + *
  9 + * This program is distributed in the hope that it will be useful,
  10 + * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11 + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  12 + * GNU General Public License for more details.
  13 + *
  14 + * You should have received a copy of the GNU General Public License
  15 + * along with this program; if not, write to the Free Software
  16 + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  17 + *
  18 + * Copyright (C) IBM Corporation, 2006
  19 + *
  20 + * Author: Ankita Garg <ankita@in.ibm.com>
  21 + *
  22 + * This module induces system failures at predefined crashpoints to
  23 + * evaluate the reliability of crash dumps obtained using different dumping
  24 + * solutions.
  25 + *
  26 + * It is adapted from the Linux Kernel Dump Test Tool by
  27 + * Fernando Luis Vazquez Cao <http://lkdtt.sourceforge.net>
  28 + *
  29 + * Usage : insmod lkdtm.ko [recur_count={>0}] cpoint_name=<> cpoint_type=<>
  30 + * [cpoint_count={>0}]
  31 + *
  32 + * recur_count : Recursion level for the stack overflow test. Default is 10.
  33 + *
  34 + * cpoint_name : Crash point where the kernel is to be crashed. It can be
  35 + * one of INT_HARDWARE_ENTRY, INT_HW_IRQ_EN, INT_TASKLET_ENTRY,
  36 + * FS_DEVRW, MEM_SWAPOUT, TIMERADD, SCSI_DISPATCH_CMD,
  37 + * IDE_CORE_CP
  38 + *
  39 + * cpoint_type : Indicates the action to be taken on hitting the crash point.
  40 + * It can be one of PANIC, BUG, EXCEPTION, LOOP, OVERFLOW
  41 + *
  42 + * cpoint_count : Indicates the number of times the crash point is to be hit
  43 + * to trigger an action. The default is 10.
  44 + */
  45 +
  46 +#include <linux/kernel.h>
  47 +#include <linux/module.h>
  48 +#include <linux/kprobes.h>
  49 +#include <linux/kallsyms.h>
  50 +#include <linux/init.h>
  51 +#include <linux/irq.h>
  52 +#include <linux/interrupt.h>
  53 +#include <scsi/scsi_cmnd.h>
  54 +
  55 +#ifdef CONFIG_IDE
  56 +#include <linux/ide.h>
  57 +#endif
  58 +
  59 +#define NUM_CPOINTS 8
  60 +#define NUM_CPOINT_TYPES 5
  61 +#define DEFAULT_COUNT 10
  62 +#define REC_NUM_DEFAULT 10
  63 +
  64 +enum cname {
  65 + INVALID,
  66 + INT_HARDWARE_ENTRY,
  67 + INT_HW_IRQ_EN,
  68 + INT_TASKLET_ENTRY,
  69 + FS_DEVRW,
  70 + MEM_SWAPOUT,
  71 + TIMERADD,
  72 + SCSI_DISPATCH_CMD,
  73 + IDE_CORE_CP
  74 +};
  75 +
  76 +enum ctype {
  77 + NONE,
  78 + PANIC,
  79 + BUG,
  80 + EXCEPTION,
  81 + LOOP,
  82 + OVERFLOW
  83 +};
  84 +
  85 +static char* cp_name[] = {
  86 + "INT_HARDWARE_ENTRY",
  87 + "INT_HW_IRQ_EN",
  88 + "INT_TASKLET_ENTRY",
  89 + "FS_DEVRW",
  90 + "MEM_SWAPOUT",
  91 + "TIMERADD",
  92 + "SCSI_DISPATCH_CMD",
  93 + "IDE_CORE_CP"
  94 +};
  95 +
  96 +static char* cp_type[] = {
  97 + "PANIC",
  98 + "BUG",
  99 + "EXCEPTION",
  100 + "LOOP",
  101 + "OVERFLOW"
  102 +};
  103 +
  104 +static struct jprobe lkdtm;
  105 +
  106 +static int lkdtm_parse_commandline(void);
  107 +static void lkdtm_handler(void);
  108 +
  109 +static char* cpoint_name = INVALID;
  110 +static char* cpoint_type = NONE;
  111 +static int cpoint_count = DEFAULT_COUNT;
  112 +static int recur_count = REC_NUM_DEFAULT;
  113 +
  114 +static enum cname cpoint = INVALID;
  115 +static enum ctype cptype = NONE;
  116 +static int count = DEFAULT_COUNT;
  117 +
  118 +module_param(recur_count, int, 0644);
  119 +MODULE_PARM_DESC(recur_count, "Recurcion level for the stack overflow test,\
  120 + default is 10");
  121 +module_param(cpoint_name, charp, 0644);
  122 +MODULE_PARM_DESC(cpoint_name, "Crash Point, where kernel is to be crashed");
  123 +module_param(cpoint_type, charp, 06444);
  124 +MODULE_PARM_DESC(cpoint_type, "Crash Point Type, action to be taken on\
  125 + hitting the crash point");
  126 +module_param(cpoint_count, int, 06444);
  127 +MODULE_PARM_DESC(cpoint_count, "Crash Point Count, number of times the \
  128 + crash point is to be hit to trigger action");
  129 +
  130 +unsigned int jp_do_irq(unsigned int irq, struct pt_regs *regs)
  131 +{
  132 + lkdtm_handler();
  133 + jprobe_return();
  134 + return 0;
  135 +}
  136 +
  137 +irqreturn_t jp_handle_irq_event(unsigned int irq, struct pt_regs *regs,
  138 + struct irqaction *action)
  139 +{
  140 + lkdtm_handler();
  141 + jprobe_return();
  142 + return 0;
  143 +}
  144 +
  145 +void jp_tasklet_action(struct softirq_action *a)
  146 +{
  147 + lkdtm_handler();
  148 + jprobe_return();
  149 +}
  150 +
  151 +void jp_ll_rw_block(int rw, int nr, struct buffer_head *bhs[])
  152 +{
  153 + lkdtm_handler();
  154 + jprobe_return();
  155 +}
  156 +
  157 +struct scan_control;
  158 +
  159 +unsigned long jp_shrink_page_list(struct list_head *page_list,
  160 + struct scan_control *sc)
  161 +{
  162 + lkdtm_handler();
  163 + jprobe_return();
  164 + return 0;
  165 +}
  166 +
  167 +int jp_hrtimer_start(struct hrtimer *timer, ktime_t tim,
  168 + const enum hrtimer_mode mode)
  169 +{
  170 + lkdtm_handler();
  171 + jprobe_return();
  172 + return 0;
  173 +}
  174 +
  175 +int jp_scsi_dispatch_cmd(struct scsi_cmnd *cmd)
  176 +{
  177 + lkdtm_handler();
  178 + jprobe_return();
  179 + return 0;
  180 +}
  181 +
  182 +#ifdef CONFIG_IDE
  183 +int jp_generic_ide_ioctl(ide_drive_t *drive, struct file *file,
  184 + struct block_device *bdev, unsigned int cmd,
  185 + unsigned long arg)
  186 +{
  187 + lkdtm_handler();
  188 + jprobe_return();
  189 + return 0;
  190 +}
  191 +#endif
  192 +
  193 +static int lkdtm_parse_commandline(void)
  194 +{
  195 + int i;
  196 +
  197 + if (cpoint_name == INVALID || cpoint_type == NONE ||
  198 + cpoint_count < 1 || recur_count < 1)
  199 + return -EINVAL;
  200 +
  201 + for (i = 0; i < NUM_CPOINTS; ++i) {
  202 + if (!strcmp(cpoint_name, cp_name[i])) {
  203 + cpoint = i + 1;
  204 + break;
  205 + }
  206 + }
  207 +
  208 + for (i = 0; i < NUM_CPOINT_TYPES; ++i) {
  209 + if (!strcmp(cpoint_type, cp_type[i])) {
  210 + cptype = i + 1;
  211 + break;
  212 + }
  213 + }
  214 +
  215 + if (cpoint == INVALID || cptype == NONE)
  216 + return -EINVAL;
  217 +
  218 + count = cpoint_count;
  219 +
  220 + return 0;
  221 +}
  222 +
  223 +static int recursive_loop(int a)
  224 +{
  225 + char buf[1024];
  226 +
  227 + memset(buf,0xFF,1024);
  228 + recur_count--;
  229 + if (!recur_count)
  230 + return 0;
  231 + else
  232 + return recursive_loop(a);
  233 +}
  234 +
  235 +void lkdtm_handler(void)
  236 +{
  237 + printk(KERN_INFO "lkdtm : Crash point %s of type %s hit\n",
  238 + cpoint_name, cpoint_type);
  239 + --count;
  240 +
  241 + if (count == 0) {
  242 + switch (cptype) {
  243 + case NONE:
  244 + break;
  245 + case PANIC:
  246 + printk(KERN_INFO "lkdtm : PANIC\n");
  247 + panic("dumptest");
  248 + break;
  249 + case BUG:
  250 + printk(KERN_INFO "lkdtm : BUG\n");
  251 + BUG();
  252 + break;
  253 + case EXCEPTION:
  254 + printk(KERN_INFO "lkdtm : EXCEPTION\n");
  255 + *((int *) 0) = 0;
  256 + break;
  257 + case LOOP:
  258 + printk(KERN_INFO "lkdtm : LOOP\n");
  259 + for (;;);
  260 + break;
  261 + case OVERFLOW:
  262 + printk(KERN_INFO "lkdtm : OVERFLOW\n");
  263 + (void) recursive_loop(0);
  264 + break;
  265 + default:
  266 + break;
  267 + }
  268 + count = cpoint_count;
  269 + }
  270 +}
  271 +
  272 +int lkdtm_module_init(void)
  273 +{
  274 + int ret;
  275 +
  276 + if (lkdtm_parse_commandline() == -EINVAL) {
  277 + printk(KERN_INFO "lkdtm : Invalid command\n");
  278 + return -EINVAL;
  279 + }
  280 +
  281 + switch (cpoint) {
  282 + case INT_HARDWARE_ENTRY:
  283 + lkdtm.kp.symbol_name = "__do_IRQ";
  284 + lkdtm.entry = (kprobe_opcode_t*) jp_do_irq;
  285 + break;
  286 + case INT_HW_IRQ_EN:
  287 + lkdtm.kp.symbol_name = "handle_IRQ_event";
  288 + lkdtm.entry = (kprobe_opcode_t*) jp_handle_irq_event;
  289 + break;
  290 + case INT_TASKLET_ENTRY:
  291 + lkdtm.kp.symbol_name = "tasklet_action";
  292 + lkdtm.entry = (kprobe_opcode_t*) jp_tasklet_action;
  293 + break;
  294 + case FS_DEVRW:
  295 + lkdtm.kp.symbol_name = "ll_rw_block";
  296 + lkdtm.entry = (kprobe_opcode_t*) jp_ll_rw_block;
  297 + break;
  298 + case MEM_SWAPOUT:
  299 + lkdtm.kp.symbol_name = "shrink_page_list";
  300 + lkdtm.entry = (kprobe_opcode_t*) jp_shrink_page_list;
  301 + break;
  302 + case TIMERADD:
  303 + lkdtm.kp.symbol_name = "hrtimer_start";
  304 + lkdtm.entry = (kprobe_opcode_t*) jp_hrtimer_start;
  305 + break;
  306 + case SCSI_DISPATCH_CMD:
  307 + lkdtm.kp.symbol_name = "scsi_dispatch_cmd";
  308 + lkdtm.entry = (kprobe_opcode_t*) jp_scsi_dispatch_cmd;
  309 + break;
  310 + case IDE_CORE_CP:
  311 +#ifdef CONFIG_IDE
  312 + lkdtm.kp.symbol_name = "generic_ide_ioctl";
  313 + lkdtm.entry = (kprobe_opcode_t*) jp_generic_ide_ioctl;
  314 +#else
  315 + printk(KERN_INFO "lkdtm : Crash point not available\n");
  316 +#endif
  317 + break;
  318 + default:
  319 + printk(KERN_INFO "lkdtm : Invalid Crash Point\n");
  320 + break;
  321 + }
  322 +
  323 + if ((ret = register_jprobe(&lkdtm)) < 0) {
  324 + printk(KERN_INFO "lkdtm : Couldn't register jprobe\n");
  325 + return ret;
  326 + }
  327 +
  328 + printk(KERN_INFO "lkdtm : Crash point %s of type %s registered\n",
  329 + cpoint_name, cpoint_type);
  330 + return 0;
  331 +}
  332 +
  333 +void lkdtm_module_exit(void)
  334 +{
  335 + unregister_jprobe(&lkdtm);
  336 + printk(KERN_INFO "lkdtm : Crash point unregistered\n");
  337 +}
  338 +
  339 +module_init(lkdtm_module_init);
  340 +module_exit(lkdtm_module_exit);
  341 +
  342 +MODULE_LICENSE("GPL");
... ... @@ -384,4 +384,18 @@
384 384 at boot time (you probably don't).
385 385 Say M if you want the RCU torture tests to build as a module.
386 386 Say N if you are unsure.
  387 +
  388 +config LKDTM
  389 + tristate "Linux Kernel Dump Test Tool Module"
  390 + depends on KPROBES
  391 + default n
  392 + help
  393 + This module enables testing of the different dumping mechanisms by
  394 + inducing system failures at predefined crash points.
  395 + If you don't need it: say N
  396 + Choose M here to compile this code as a module. The module will be
  397 + called lkdtm.
  398 +
  399 + Documentation on how to use the module can be found in
  400 + drivers/misc/lkdtm.c