1
linux/drivers/misc/lkdtm.c

344 lines
7.7 KiB
C
Raw Normal View History

/*
* Kprobe module for testing crash dumps
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
*
* Copyright (C) IBM Corporation, 2006
*
* Author: Ankita Garg <ankita@in.ibm.com>
*
* This module induces system failures at predefined crashpoints to
* evaluate the reliability of crash dumps obtained using different dumping
* solutions.
*
* It is adapted from the Linux Kernel Dump Test Tool by
* Fernando Luis Vazquez Cao <http://lkdtt.sourceforge.net>
*
* Usage : insmod lkdtm.ko [recur_count={>0}] cpoint_name=<> cpoint_type=<>
* [cpoint_count={>0}]
*
* recur_count : Recursion level for the stack overflow test. Default is 10.
*
* cpoint_name : Crash point where the kernel is to be crashed. It can be
* one of INT_HARDWARE_ENTRY, INT_HW_IRQ_EN, INT_TASKLET_ENTRY,
* FS_DEVRW, MEM_SWAPOUT, TIMERADD, SCSI_DISPATCH_CMD,
* IDE_CORE_CP
*
* cpoint_type : Indicates the action to be taken on hitting the crash point.
* It can be one of PANIC, BUG, EXCEPTION, LOOP, OVERFLOW
*
* cpoint_count : Indicates the number of times the crash point is to be hit
* to trigger an action. The default is 10.
*/
#include <linux/kernel.h>
#include <linux/fs.h>
#include <linux/module.h>
#include <linux/buffer_head.h>
#include <linux/kprobes.h>
#include <linux/list.h>
#include <linux/init.h>
#include <linux/interrupt.h>
#include <linux/hrtimer.h>
#include <scsi/scsi_cmnd.h>
#ifdef CONFIG_IDE
#include <linux/ide.h>
#endif
#define NUM_CPOINTS 8
#define NUM_CPOINT_TYPES 5
#define DEFAULT_COUNT 10
#define REC_NUM_DEFAULT 10
enum cname {
INVALID,
INT_HARDWARE_ENTRY,
INT_HW_IRQ_EN,
INT_TASKLET_ENTRY,
FS_DEVRW,
MEM_SWAPOUT,
TIMERADD,
SCSI_DISPATCH_CMD,
IDE_CORE_CP
};
enum ctype {
NONE,
PANIC,
BUG,
EXCEPTION,
LOOP,
OVERFLOW
};
static char* cp_name[] = {
"INT_HARDWARE_ENTRY",
"INT_HW_IRQ_EN",
"INT_TASKLET_ENTRY",
"FS_DEVRW",
"MEM_SWAPOUT",
"TIMERADD",
"SCSI_DISPATCH_CMD",
"IDE_CORE_CP"
};
static char* cp_type[] = {
"PANIC",
"BUG",
"EXCEPTION",
"LOOP",
"OVERFLOW"
};
static struct jprobe lkdtm;
static int lkdtm_parse_commandline(void);
static void lkdtm_handler(void);
static char* cpoint_name = INVALID;
static char* cpoint_type = NONE;
static int cpoint_count = DEFAULT_COUNT;
static int recur_count = REC_NUM_DEFAULT;
static enum cname cpoint = INVALID;
static enum ctype cptype = NONE;
static int count = DEFAULT_COUNT;
module_param(recur_count, int, 0644);
MODULE_PARM_DESC(recur_count, " Recursion level for the stack overflow test, "\
"default is 10");
module_param(cpoint_name, charp, 0644);
MODULE_PARM_DESC(cpoint_name, " Crash Point, where kernel is to be crashed");
module_param(cpoint_type, charp, 0644);
MODULE_PARM_DESC(cpoint_type, " Crash Point Type, action to be taken on "\
"hitting the crash point");
module_param(cpoint_count, int, 0644);
MODULE_PARM_DESC(cpoint_count, " Crash Point Count, number of times the "\
"crash point is to be hit to trigger action");
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 06:55:46 -07:00
unsigned int jp_do_irq(unsigned int irq)
{
lkdtm_handler();
jprobe_return();
return 0;
}
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 06:55:46 -07:00
irqreturn_t jp_handle_irq_event(unsigned int irq, struct irqaction *action)
{
lkdtm_handler();
jprobe_return();
return 0;
}
void jp_tasklet_action(struct softirq_action *a)
{
lkdtm_handler();
jprobe_return();
}
void jp_ll_rw_block(int rw, int nr, struct buffer_head *bhs[])
{
lkdtm_handler();
jprobe_return();
}
struct scan_control;
unsigned long jp_shrink_inactive_list(unsigned long max_scan,
struct zone *zone, struct scan_control *sc)
{
lkdtm_handler();
jprobe_return();
return 0;
}
int jp_hrtimer_start(struct hrtimer *timer, ktime_t tim,
const enum hrtimer_mode mode)
{
lkdtm_handler();
jprobe_return();
return 0;
}
int jp_scsi_dispatch_cmd(struct scsi_cmnd *cmd)
{
lkdtm_handler();
jprobe_return();
return 0;
}
#ifdef CONFIG_IDE
int jp_generic_ide_ioctl(ide_drive_t *drive, struct file *file,
struct block_device *bdev, unsigned int cmd,
unsigned long arg)
{
lkdtm_handler();
jprobe_return();
return 0;
}
#endif
static int lkdtm_parse_commandline(void)
{
int i;
if (cpoint_name == INVALID || cpoint_type == NONE ||
cpoint_count < 1 || recur_count < 1)
return -EINVAL;
for (i = 0; i < NUM_CPOINTS; ++i) {
if (!strcmp(cpoint_name, cp_name[i])) {
cpoint = i + 1;
break;
}
}
for (i = 0; i < NUM_CPOINT_TYPES; ++i) {
if (!strcmp(cpoint_type, cp_type[i])) {
cptype = i + 1;
break;
}
}
if (cpoint == INVALID || cptype == NONE)
return -EINVAL;
count = cpoint_count;
return 0;
}
static int recursive_loop(int a)
{
char buf[1024];
memset(buf,0xFF,1024);
recur_count--;
if (!recur_count)
return 0;
else
return recursive_loop(a);
}
void lkdtm_handler(void)
{
printk(KERN_INFO "lkdtm : Crash point %s of type %s hit\n",
cpoint_name, cpoint_type);
--count;
if (count == 0) {
switch (cptype) {
case NONE:
break;
case PANIC:
printk(KERN_INFO "lkdtm : PANIC\n");
panic("dumptest");
break;
case BUG:
printk(KERN_INFO "lkdtm : BUG\n");
BUG();
break;
case EXCEPTION:
printk(KERN_INFO "lkdtm : EXCEPTION\n");
*((int *) 0) = 0;
break;
case LOOP:
printk(KERN_INFO "lkdtm : LOOP\n");
for (;;);
break;
case OVERFLOW:
printk(KERN_INFO "lkdtm : OVERFLOW\n");
(void) recursive_loop(0);
break;
default:
break;
}
count = cpoint_count;
}
}
int lkdtm_module_init(void)
{
int ret;
if (lkdtm_parse_commandline() == -EINVAL) {
printk(KERN_INFO "lkdtm : Invalid command\n");
return -EINVAL;
}
switch (cpoint) {
case INT_HARDWARE_ENTRY:
lkdtm.kp.symbol_name = "__do_IRQ";
lkdtm.entry = (kprobe_opcode_t*) jp_do_irq;
break;
case INT_HW_IRQ_EN:
lkdtm.kp.symbol_name = "handle_IRQ_event";
lkdtm.entry = (kprobe_opcode_t*) jp_handle_irq_event;
break;
case INT_TASKLET_ENTRY:
lkdtm.kp.symbol_name = "tasklet_action";
lkdtm.entry = (kprobe_opcode_t*) jp_tasklet_action;
break;
case FS_DEVRW:
lkdtm.kp.symbol_name = "ll_rw_block";
lkdtm.entry = (kprobe_opcode_t*) jp_ll_rw_block;
break;
case MEM_SWAPOUT:
lkdtm.kp.symbol_name = "shrink_inactive_list";
lkdtm.entry = (kprobe_opcode_t*) jp_shrink_inactive_list;
break;
case TIMERADD:
lkdtm.kp.symbol_name = "hrtimer_start";
lkdtm.entry = (kprobe_opcode_t*) jp_hrtimer_start;
break;
case SCSI_DISPATCH_CMD:
lkdtm.kp.symbol_name = "scsi_dispatch_cmd";
lkdtm.entry = (kprobe_opcode_t*) jp_scsi_dispatch_cmd;
break;
case IDE_CORE_CP:
#ifdef CONFIG_IDE
lkdtm.kp.symbol_name = "generic_ide_ioctl";
lkdtm.entry = (kprobe_opcode_t*) jp_generic_ide_ioctl;
#else
printk(KERN_INFO "lkdtm : Crash point not available\n");
#endif
break;
default:
printk(KERN_INFO "lkdtm : Invalid Crash Point\n");
break;
}
if ((ret = register_jprobe(&lkdtm)) < 0) {
printk(KERN_INFO "lkdtm : Couldn't register jprobe\n");
return ret;
}
printk(KERN_INFO "lkdtm : Crash point %s of type %s registered\n",
cpoint_name, cpoint_type);
return 0;
}
void lkdtm_module_exit(void)
{
unregister_jprobe(&lkdtm);
printk(KERN_INFO "lkdtm : Crash point unregistered\n");
}
module_init(lkdtm_module_init);
module_exit(lkdtm_module_exit);
MODULE_LICENSE("GPL");