kprobes.txt

Title	: Kernel Probes (Kprobes)
Authors	: Jim Keniston <jkenisto@us.ibm.com>
	: Prasanna S Panchamukhi <prasanna@in.ibm.com>

CONTENTS

1. Concepts: Kprobes, Jprobes, Return Probes
2. Architectures Supported
3. Configuring Kprobes
4. API Reference
5. Kprobes Features and Limitations
6. Probe Overhead
7. TODO
8. Kprobes Example
9. Jprobes Example
10. Kretprobes Example

1. Concepts: Kprobes, Jprobes, Return Probes

Kprobes enables you to dynamically break into any kernel routine and
collect debugging and performance information non-disruptively. You
can trap at almost any kernel code address, specifying a handler
routine to be invoked when the breakpoint is hit.

There are currently three types of probes: kprobes, jprobes, and
kretprobes (also called return probes).  A kprobe can be inserted
on virtually any instruction in the kernel.  A jprobe is inserted at
the entry to a kernel function, and provides convenient access to the
function's arguments.  A return probe fires when a specified function
returns.

In the typical case, Kprobes-based instrumentation is packaged as
a kernel module.  The module's init function installs ("registers")
one or more probes, and the exit function unregisters them.  A
registration function such as register_kprobe() specifies where
the probe is to be inserted and what handler is to be called when
the probe is hit.

The next three subsections explain how the different types of
probes work.  They explain certain things that you'll need to
know in order to make the best use of Kprobes -- e.g., the
difference between a pre_handler and a post_handler, and how
to use the maxactive and nmissed fields of a kretprobe.  But
if you're in a hurry to start using Kprobes, you can skip ahead
to section 2.

1.1 How Does a Kprobe Work?

When a kprobe is registered, Kprobes makes a copy of the probed
instruction and replaces the first byte(s) of the probed instruction
with a breakpoint instruction (e.g., int3 on i386 and x86_64).

When a CPU hits the breakpoint instruction, a trap occurs, the CPU's
registers are saved, and control passes to Kprobes via the
notifier_call_chain mechanism.  Kprobes executes the "pre_handler"
associated with the kprobe, passing the handler the addresses of the
kprobe struct and the saved registers.

Next, Kprobes single-steps its copy of the probed instruction.
(It would be simpler to single-step the actual instruction in place,
but then Kprobes would have to temporarily remove the breakpoint
instruction.  This would open a small time window when another CPU
could sail right past the probepoint.)

After the instruction is single-stepped, Kprobes executes the
"post_handler," if any, that is associated with the kprobe.
Execution then continues with the instruction following the probepoint.

1.2 How Does a Jprobe Work?

A jprobe is implemented using a kprobe that is placed on a function's
entry point.  It employs a simple mirroring principle to allow
seamless access to the probed function's arguments.  The jprobe
handler routine should have the same signature (arg list and return
type) as the function being probed, and must always end by calling
the Kprobes function jprobe_return().

Here's how it works.  When the probe is hit, Kprobes makes a copy of
the saved registers and a generous portion of the stack (see below).
Kprobes then points the saved instruction pointer at the jprobe's
handler routine, and returns from the trap.  As a result, control
passes to the handler, which is presented with the same register and
stack contents as the probed function.  When it is done, the handler
calls jprobe_return(), which traps again to restore the original stack
contents and processor state and switch to the probed function.

By convention, the callee owns its arguments, so gcc may produce code
that unexpectedly modifies that portion of the stack.  This is why
Kprobes saves a copy of the stack and restores it after the jprobe
handler has run.  Up to MAX_STACK_SIZE bytes are copied -- e.g.,
64 bytes on i386.

Note that the probed function's args may be passed on the stack
or in registers (e.g., for x86_64 or for an i386 fastcall function).
The jprobe will work in either case, so long as the handler's
prototype matches that of the probed function.

1.3 How Does a Return Probe Work?

When you call register_kretprobe(), Kprobes establishes a kprobe at
the entry to the function.  When the probed function is called and this
probe is hit, Kprobes saves a copy of the return address, and replaces
the return address with the address of a "trampoline."  The trampoline
is an arbitrary piece of code -- typically just a nop instruction.
At boot time, Kprobes registers a kprobe at the trampoline.

When the probed function executes its return instruction, control
passes to the trampoline and that probe is hit.  Kprobes' trampoline
handler calls the user-specified handler associated with the kretprobe,
then sets the saved instruction pointer to the saved return address,
and that's where execution resumes upon return from the trap.

While the probed function is executing, its return address is
stored in an object of type kretprobe_instance.  Before calling
register_kretprobe(), the user sets the maxactive field of the
kretprobe struct to specify how many instances of the specified
function can be probed simultaneously.  register_kretprobe()
pre-allocates the indicated number of kretprobe_instance objects.

For example, if the function is non-recursive and is called with a
spinlock held, maxactive = 1 should be enough.  If the function is
non-recursive and can never relinquish the CPU (e.g., via a semaphore
or preemption), NR_CPUS should be enough.  If maxactive <= 0, it is
set to a default value.  If CONFIG_PREEMPT is enabled, the default
is max(10, 2*NR_CPUS).  Otherwise, the default is NR_CPUS.

It's not a disaster if you set maxactive too low; you'll just miss
some probes.  In the kretprobe struct, the nmissed field is set to
zero when the return probe is registered, and is incremented every
time the probed function is entered but there is no kretprobe_instance
object available for establishing the return probe.

2. Architectures Supported

Kprobes, jprobes, and return probes are implemented on the following
architectures:

- i386
- x86_64 (AMD-64, EM64T)
- ppc64
- ia64 (Does not support probes on instruction slot1.)
- sparc64 (Return probes not yet implemented.)

3. Configuring Kprobes

When configuring the kernel using make menuconfig/xconfig/oldconfig,
ensure that CONFIG_KPROBES is set to "y".  Under "Instrumentation
Support", look for "Kprobes".

So that you can load and unload Kprobes-based instrumentation modules,
make sure "Loadable module support" (CONFIG_MODULES) and "Module
unloading" (CONFIG_MODULE_UNLOAD) are set to "y".

You may also want to ensure that CONFIG_KALLSYMS and perhaps even
CONFIG_KALLSYMS_ALL are set to "y", since kallsyms_lookup_name()
is a handy, version-independent way to find a function's address.

If you need to insert a probe in the middle of a function, you may find
it useful to "Compile the kernel with debug info" (CONFIG_DEBUG_INFO),
so you can use "objdump -d -l vmlinux" to see the source-to-object
code mapping.

4. API Reference

The Kprobes API includes a "register" function and an "unregister"
function for each type of probe.  Here are terse, mini-man-page
specifications for these functions and the associated probe handlers
that you'll write.  See the latter half of this document for examples.

4.1 register_kprobe

#include <linux/kprobes.h>
int register_kprobe(struct kprobe *kp);

Sets a breakpoint at the address kp->addr.  When the breakpoint is
hit, Kprobes calls kp->pre_handler.  After the probed instruction
is single-stepped, Kprobe calls kp->post_handler.  If a fault
occurs during execution of kp->pre_handler or kp->post_handler,
or during single-stepping of the probed instruction, Kprobes calls
kp->fault_handler.  Any or all handlers can be NULL.

register_kprobe() returns 0 on success, or a negative errno otherwise.

User's pre-handler (kp->pre_handler):
#include <linux/kprobes.h>
#include <linux/ptrace.h>
int pre_handler(struct kprobe *p, struct pt_regs *regs);

Called with p pointing to the kprobe associated with the breakpoint,
and regs pointing to the struct containing the registers saved when
the breakpoint was hit.  Return 0 here unless you're a Kprobes geek.

User's post-handler (kp->post_handler):
#include <linux/kprobes.h>
#include <linux/ptrace.h>
void post_handler(struct kprobe *p, struct pt_regs *regs,
	unsigned long flags);

p and regs are as described for the pre_handler.  flags always seems
to be zero.

User's fault-handler (kp->fault_handler):
#include <linux/kprobes.h>
#include <linux/ptrace.h>
int fault_handler(struct kprobe *p, struct pt_regs *regs, int trapnr);

p and regs are as described for the pre_handler.  trapnr is the
architecture-specific trap number associated with the fault (e.g.,
on i386, 13 for a general protection fault or 14 for a page fault).
Returns 1 if it successfully handled the exception.

4.2 register_jprobe

#include <linux/kprobes.h>
int register_jprobe(struct jprobe *jp)

Sets a breakpoint at the address jp->kp.addr, which must be the address
of the first instruction of a function.  When the breakpoint is hit,
Kprobes runs the handler whose address is jp->entry.

The handler should have the same arg list and return type as the probed
function; and just before it returns, it must call jprobe_return().
(The handler never actually returns, since jprobe_return() returns
control to Kprobes.)  If the probed function is declared asmlinkage,
fastcall, or anything else that affects how args are passed, the
handler's declaration must match.

register_jprobe() returns 0 on success, or a negative errno otherwise.

4.3 register_kretprobe

#include <linux/kprobes.h>
int register_kretprobe(struct kretprobe *rp);

Establishes a return probe for the function whose address is
rp->kp.addr.  When that function returns, Kprobes calls rp->handler.
You must set rp->maxactive appropriately before you call
register_kretprobe(); see "How Does a Return Probe Work?" for details.

register_kretprobe() returns 0 on success, or a negative errno
otherwise.

User's return-probe handler (rp->handler):
#include <linux/kprobes.h>
#include <linux/ptrace.h>
int kretprobe_handler(struct kretprobe_instance *ri, struct pt_regs *regs);

regs is as described for kprobe.pre_handler.  ri points to the
kretprobe_instance object, of which the following fields may be
of interest:
- ret_addr: the return address
- rp: points to the corresponding kretprobe object
- task: points to the corresponding task struct
The handler's return value is currently ignored.

4.4 unregister_*probe

#include <linux/kprobes.h>
void unregister_kprobe(struct kprobe *kp);
void unregister_jprobe(struct jprobe *jp);
void unregister_kretprobe(struct kretprobe *rp);

Removes the specified probe.  The unregister function can be called
at any time after the probe has been registered.

5. Kprobes Features and Limitations

Kprobes allows multiple probes at the same address.  Currently,
however, there cannot be multiple jprobes on the same function at
the same time.

In general, you can install a probe anywhere in the kernel.
In particular, you can probe interrupt handlers.  Known exceptions
are discussed in this section.

The register_*probe functions will return -EINVAL if you attempt
to install a probe in the code that implements Kprobes (mostly
kernel/kprobes.c and arch/*/kernel/kprobes.c, but also functions such
as do_page_fault and notifier_call_chain).

If you install a probe in an inline-able function, Kprobes makes
no attempt to chase down all inline instances of the function and
install probes there.  gcc may inline a function without being asked,
so keep this in mind if you're not seeing the probe hits you expect.

A probe handler can modify the environment of the probed function
-- e.g., by modifying kernel data structures, or by modifying the
contents of the pt_regs struct (which are restored to the registers
upon return from the breakpoint).  So Kprobes can be used, for example,
to install a bug fix or to inject faults for testing.  Kprobes, of
course, has no way to distinguish the deliberately injected faults
from the accidental ones.  Don't drink and probe.

Kprobes makes no attempt to prevent probe handlers from stepping on
each other -- e.g., probing printk() and then calling printk() from a
probe handler.  If a probe handler hits a probe, that second probe's
handlers won't be run in that instance, and the kprobe.nmissed member
of the second probe will be incremented.

As of Linux v2.6.15-rc1, multiple handlers (or multiple instances of
the same handler) may run concurrently on different CPUs.

Kprobes does not use mutexes or allocate memory except during
registration and unregistration.

Probe handlers are run with preemption disabled.  Depending on the
architecture, handlers may also run with interrupts disabled.  In any
case, your handler should not yield the CPU (e.g., by attempting to
acquire a semaphore).

Since a return probe is implemented by replacing the return
address with the trampoline's address, stack backtraces and calls
to __builtin_return_address() will typically yield the trampoline's
address instead of the real return address for kretprobed functions.
(As far as we can tell, __builtin_return_address() is used only
for instrumentation and error reporting.)

If the number of times a function is called does not match the number
of times it returns, registering a return probe on that function may
produce undesirable results.  We have the do_exit() case covered.
do_execve() and do_fork() are not an issue.  We're unaware of other
specific cases where this could be a problem.

If, upon entry to or exit from a function, the CPU is running on
a stack other than that of the current task, registering a return
probe on that function may produce undesirable results.  For this
reason, Kprobes doesn't support return probes (or kprobes or jprobes)
on the x86_64 version of __switch_to(); the registration functions
return -EINVAL.

6. Probe Overhead

On a typical CPU in use in 2005, a kprobe hit takes 0.5 to 1.0
microseconds to process.  Specifically, a benchmark that hits the same
probepoint repeatedly, firing a simple handler each time, reports 1-2
million hits per second, depending on the architecture.  A jprobe or
return-probe hit typically takes 50-75% longer than a kprobe hit.
When you have a return probe set on a function, adding a kprobe at
the entry to that function adds essentially no overhead.

Here are sample overhead figures (in usec) for different architectures.
k = kprobe; j = jprobe; r = return probe; kr = kprobe + return probe
on same function; jr = jprobe + return probe on same function

i386: Intel Pentium M, 1495 MHz, 2957.31 bogomips
k = 0.57 usec; j = 1.00; r = 0.92; kr = 0.99; jr = 1.40

x86_64: AMD Opteron 246, 1994 MHz, 3971.48 bogomips
k = 0.49 usec; j = 0.76; r = 0.80; kr = 0.82; jr = 1.07

ppc64: POWER5 (gr), 1656 MHz (SMT disabled, 1 virtual CPU per physical CPU)
k = 0.77 usec; j = 1.31; r = 1.26; kr = 1.45; jr = 1.99

7. TODO

a. SystemTap (http://sourceware.org/systemtap): Provides a simplified
programming interface for probe-based instrumentation.  Try it out.
b. Kernel return probes for sparc64.
c. Support for other architectures.
d. User-space probes.
e. Watchpoint probes (which fire on data references).

8. Kprobes Example

Here's a sample kernel module showing the use of kprobes to dump a
stack trace and selected i386 registers when do_fork() is called.
----- cut here -----
/*kprobe_example.c*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/kprobes.h>
#include <linux/kallsyms.h>
#include <linux/sched.h>

/*For each probe you need to allocate a kprobe structure*/
static struct kprobe kp;

/*kprobe pre_handler: called just before the probed instruction is executed*/
int handler_pre(struct kprobe *p, struct pt_regs *regs)
{
	printk("pre_handler: p->addr=0x%p, eip=%lx, eflags=0x%lx\n",
		p->addr, regs->eip, regs->eflags);
	dump_stack();
	return 0;
}

/*kprobe post_handler: called after the probed instruction is executed*/
void handler_post(struct kprobe *p, struct pt_regs *regs, unsigned long flags)
{
	printk("post_handler: p->addr=0x%p, eflags=0x%lx\n",
		p->addr, regs->eflags);
}

/* fault_handler: this is called if an exception is generated for any
 * instruction within the pre- or post-handler, or when Kprobes
 * single-steps the probed instruction.
 */
int handler_fault(struct kprobe *p, struct pt_regs *regs, int trapnr)
{
	printk("fault_handler: p->addr=0x%p, trap #%dn",
		p->addr, trapnr);
	/* Return 0 because we don't handle the fault. */
	return 0;
}

int init_module(void)
{
	int ret;
	kp.pre_handler = handler_pre;
	kp.post_handler = handler_post;
	kp.fault_handler = handler_fault;
	kp.addr = (kprobe_opcode_t*) kallsyms_lookup_name("do_fork");
	/* register the kprobe now */
	if (!kp.addr) {
		printk("Couldn't find %s to plant kprobe\n", "do_fork");
		return -1;
	}
	if ((ret = register_kprobe(&kp) < 0)) {
		printk("register_kprobe failed, returned %d\n", ret);
		return -1;
	}
	printk("kprobe registered\n");
	return 0;
}

void cleanup_module(void)
{
	unregister_kprobe(&kp);
	printk("kprobe unregistered\n");
}

MODULE_LICENSE("GPL");
----- cut here -----

You can build the kernel module, kprobe-example.ko, using the following
Makefile:
----- cut here -----
obj-m := kprobe-example.o
KDIR := /lib/modules/$(shell uname -r)/build
PWD := $(shell pwd)
default:
	$(MAKE) -C $(KDIR) SUBDIRS=$(PWD) modules
clean:
	rm -f *.mod.c *.ko *.o
----- cut here -----

$ make
$ su -
...
# insmod kprobe-example.ko

You will see the trace data in /var/log/messages and on the console
whenever do_fork() is invoked to create a new process.

9. Jprobes Example

Here's a sample kernel module showing the use of jprobes to dump
the arguments of do_fork().
----- cut here -----
/*jprobe-example.c */
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/fs.h>
#include <linux/uio.h>
#include <linux/kprobes.h>
#include <linux/kallsyms.h>

/*
 * Jumper probe for do_fork.
 * Mirror principle enables access to arguments of the probed routine
 * from the probe handler.
 */

/* Proxy routine having the same arguments as actual do_fork() routine */
long jdo_fork(unsigned long clone_flags, unsigned long stack_start,
	      struct pt_regs *regs, unsigned long stack_size,
	      int __user * parent_tidptr, int __user * child_tidptr)
{
	printk("jprobe: clone_flags=0x%lx, stack_size=0x%lx, regs=0x%p\n",
	       clone_flags, stack_size, regs);
	/* Always end with a call to jprobe_return(). */
	jprobe_return();
	/*NOTREACHED*/
	return 0;
}

static struct jprobe my_jprobe = {
	.entry = (kprobe_opcode_t *) jdo_fork
};

int init_module(void)
{
	int ret;
	my_jprobe.kp.addr = (kprobe_opcode_t *) kallsyms_lookup_name("do_fork");
	if (!my_jprobe.kp.addr) {
		printk("Couldn't find %s to plant jprobe\n", "do_fork");
		return -1;
	}

	if ((ret = register_jprobe(&my_jprobe)) <0) {
		printk("register_jprobe failed, returned %d\n", ret);
		return -1;
	}
	printk("Planted jprobe at %p, handler addr %p\n",
	       my_jprobe.kp.addr, my_jprobe.entry);
	return 0;
}

void cleanup_module(void)
{
	unregister_jprobe(&my_jprobe);
	printk("jprobe unregistered\n");
}

MODULE_LICENSE("GPL");
----- cut here -----

Build and insert the kernel module as shown in the above kprobe
example.  You will see the trace data in /var/log/messages and on
the console whenever do_fork() is invoked to create a new process.
(Some messages may be suppressed if syslogd is configured to
eliminate duplicate messages.)

10. Kretprobes Example

Here's a sample kernel module showing the use of return probes to
report failed calls to sys_open().
----- cut here -----
/*kretprobe-example.c*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/kprobes.h>
#include <linux/kallsyms.h>

static const char *probed_func = "sys_open";

/* Return-probe handler: If the probed function fails, log the return value. */
static int ret_handler(struct kretprobe_instance *ri, struct pt_regs *regs)
{
	// Substitute the appropriate register name for your architecture --
	// e.g., regs->rax for x86_64, regs->gpr[3] for ppc64.
	int retval = (int) regs->eax;
	if (retval < 0) {
		printk("%s returns %d\n", probed_func, retval);
	}
	return 0;
}

static struct kretprobe my_kretprobe = {
	.handler = ret_handler,
	/* Probe up to 20 instances concurrently. */
	.maxactive = 20
};

int init_module(void)
{
	int ret;
	my_kretprobe.kp.addr =
		(kprobe_opcode_t *) kallsyms_lookup_name(probed_func);
	if (!my_kretprobe.kp.addr) {
		printk("Couldn't find %s to plant return probe\n", probed_func);
		return -1;
	}
	if ((ret = register_kretprobe(&my_kretprobe)) < 0) {
		printk("register_kretprobe failed, returned %d\n", ret);
		return -1;
	}
	printk("Planted return probe at %p\n", my_kretprobe.kp.addr);
	return 0;
}

void cleanup_module(void)
{
	unregister_kretprobe(&my_kretprobe);
	printk("kretprobe unregistered\n");
	/* nmissed > 0 suggests that maxactive was set too low. */
	printk("Missed probing %d instances of %s\n",
		my_kretprobe.nmissed, probed_func);
}

MODULE_LICENSE("GPL");
----- cut here -----

Build and insert the kernel module as shown in the above kprobe
example.  You will see the trace data in /var/log/messages and on the
console whenever sys_open() returns a negative value.  (Some messages
may be suppressed if syslogd is configured to eliminate duplicate
messages.)

For additional information on Kprobes, refer to the following URLs:
http://www-106.ibm.com/developerworks/library/l-kprobes.html?ca=dgr-lnxw42Kprobe
http://www.redhat.com/magazine/005mar05/features/kprobes/