Kernel Probes (Kprobes)

Kprobes와 return probe 개념

이 문서는 Jim Keniston, Prasanna S Panchamukhi, Masami Hiramatsu가 작성했다. Kprobes는 실행 중인 kernel routine의 거의 모든 code address에 동적으로 breakpoint를 설치하고, system 동작을 방해하지 않으면서 debugging 및 performance 정보를 수집하게 한다. 다만 blacklist에 속한 일부 kernel code에는 probe를 설치할 수 없다.

Probe 종류는 일반 `kprobe`와 return probe인 `kretprobe` 두 가지다. Kprobe는 kernel의 거의 모든 instruction에 삽입할 수 있고, kretprobe는 지정한 function이 return할 때 동작한다.

일반적인 instrumentation은 kernel module로 묶는다. Module init function이 `register_kprobe()` 같은 API로 probe 위치와 handler를 등록하고 exit function이 해제한다. 많은 probe를 한꺼번에 처리할 때는 `register_/unregister_*probes()` batch API가 해제 시간을 줄일 수 있다.

뒤 절은 probe별 동작과 jump optimization, `pre_handler`와 `post_handler`의 차이, kretprobe의 `maxactive`와 `nmissed` 사용법을 설명한다.

=======================
Kernel Probes (Kprobes)
=======================

:Author: Jim Keniston <[email protected]>
:Author: Prasanna S Panchamukhi <[email protected]>
:Author: Masami Hiramatsu <[email protected]>

.. CONTENTS

  1. Concepts: Kprobes, and 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. Kretprobes Example
  10. Deprecated Features
  Appendix A: The kprobes debugfs interface
  Appendix B: The kprobes sysctl interface
  Appendix C: References

Concepts: Kprobes and 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 [1]_, specifying a handler
routine to be invoked when the breakpoint is hit.

.. [1] some parts of the kernel code can not be trapped, see
       :ref:`kprobes_blacklist`)

There are currently two types of probes: kprobes, and kretprobes
(also called return probes).  A kprobe can be inserted on virtually
any instruction in the kernel.  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.

There are also ``register_/unregister_*probes()`` functions for batch
registration/unregistration of a group of ``*probes``. These functions
can speed up unregistration process when you have to unregister
a lot of probes at once.

The next four subsections explain how the different types of
probes work and how jump optimization works.  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 :ref:`kprobes_archs_supported`.

Kprobe 동작과 실행 경로 변경

Kprobe를 등록하면 Kprobes는 대상 instruction을 복사하고 원래 instruction의 첫 byte들을 breakpoint instruction으로 바꾼다. i386과 x86_64에서는 `int3`가 대표적이다.

CPU가 breakpoint를 만나면 trap이 발생하고 register가 저장된다. 제어는 `notifier_call_chain`을 통해 Kprobes로 넘어가며, Kprobes는 kprobe struct와 저장 register의 address를 넘겨 `pre_handler`를 실행한다.

그다음 원래 위치가 아니라 복사한 instruction을 single-step한다. 원래 instruction을 일시 복원하는 방식은 다른 CPU가 그 짧은 틈에 probepoint를 지나칠 수 있기 때문이다. Single-step 뒤 `post_handler`가 있으면 실행하고 probepoint 다음 instruction으로 돌아간다.

`pre_handler`는 register set과 instruction pointer를 바꿔 실행 경로를 변경할 수 있지만 stack frame과 복귀 경로를 정확히 유지해야 하므로 매우 위험하다. Instruction pointer를 바꾸고 관련 register를 설정했다면 0이 아닌 값을 반환해 Kprobes가 single-step을 중단하고 지정 address로 바로 복귀하게 해야 한다. 이 경우 `post_handler`도 호출되지 않는다.

Function call에 TOC(Table of Contents)를 사용하는 architecture에서는 module의 새 function에 맞는 TOC를 설정하고 return 뒤 이전 TOC를 복원해야 하므로 더 어렵다.

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.

Changing Execution Path
-----------------------

Since kprobes can probe into a running kernel code, it can change the
register set, including instruction pointer. This operation requires
maximum care, such as keeping the stack frame, recovering the execution
path etc. Since it operates on a running kernel and needs deep knowledge
of computer architecture and concurrent computing, you can easily shoot
your foot.

If you change the instruction pointer (and set up other related
registers) in pre_handler, you must return !0 so that kprobes stops
single stepping and just returns to the given address.
This also means post_handler should not be called anymore.

Note that this operation may be harder on some architectures which use
TOC (Table of Contents) for function call, since you have to setup a new
TOC for your function in your module, and recover the old one after
returning from it.

Return probe와 entry handler

`register_kretprobe()`는 대상 function entry에 kprobe를 설치한다. Function이 호출되면 return address를 저장하고 이를 `trampoline` address로 바꾼다. Boot 때 Kprobes는 보통 `nop`인 trampoline에도 kprobe를 등록한다.

대상 function이 return instruction을 실행하면 trampoline probe가 적중한다. Trampoline handler는 사용자가 지정한 return handler를 호출하고 저장 instruction pointer를 원래 return address로 바꾼 뒤 그 위치에서 실행을 재개한다.

Function 실행 중 return address는 `kretprobe_instance`에 저장된다. 등록 전 `kretprobe.maxactive`로 동시에 추적할 function instance 수를 정하며, `register_kretprobe()`가 그 수만큼 instance object를 미리 할당한다.

Non-recursive function이 spinlock을 잡은 채 호출되면 `maxactive=1`이면 충분하다. CPU를 절대 양보하지 않는 non-recursive function은 `NR_CPUS`면 충분하다. `maxactive <= 0`이면 기본값 `max(10, 2*NR_CPUS)`를 사용한다.

`maxactive`가 너무 작아도 치명적이지는 않지만 일부 probe를 놓친다. 등록 시 0인 `nmissed`는 function entry에서 사용할 `kretprobe_instance`가 없을 때마다 증가한다.

선택적 `entry_handler`는 function entry에서 실행된다. 0을 반환하면 해당 instance의 return 때 return handler 호출이 보장된다. 0이 아닌 error를 반환하면 return address를 바꾸지 않아 그 function instance에는 kretprobe가 더 이상 영향을 주지 않는다.

Entry와 return handler 호출은 고유 `kretprobe_instance`로 대응된다. `kretprobe.data_size`를 지정하면 instance마다 private data를 둘 수 있고 handler는 `kretprobe_instance.data`로 공유한다. 사용할 instance가 없어 `nmissed`가 증가하는 경우 `entry_handler`도 건너뛴다.

Return Probes
-------------

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 return 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: max(10, 2*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.

Kretprobe entry-handler
^^^^^^^^^^^^^^^^^^^^^^^

Kretprobes also provides an optional user-specified handler which runs
on function entry. This handler is specified by setting the entry_handler
field of the kretprobe struct. Whenever the kprobe placed by kretprobe at the
function entry is hit, the user-defined entry_handler, if any, is invoked.
If the entry_handler returns 0 (success) then a corresponding return handler
is guaranteed to be called upon function return. If the entry_handler
returns a non-zero error then Kprobes leaves the return address as is, and
the kretprobe has no further effect for that particular function instance.

Multiple entry and return handler invocations are matched using the unique
kretprobe_instance object associated with them. Additionally, a user
may also specify per return-instance private data to be part of each
kretprobe_instance object. This is especially useful when sharing private
data between corresponding user entry and return handlers. The size of each
private data object can be specified at kretprobe registration time by
setting the data_size field of the kretprobe struct. This data can be
accessed through the data field of each kretprobe_instance object.

In case probed function is entered but there is no kretprobe_instance
object available, then in addition to incrementing the nmissed count,
the user entry_handler invocation is also skipped.

Jump optimization

`CONFIG_OPTPROBES=y`이고 `debug.kprobes_optimization=1`이면 Kprobes는 각 probepoint의 breakpoint 대신 jump instruction을 사용해 probe hit overhead를 줄이려 한다. 현재 x86/x86-64 non-preemptive kernel에서는 이 config가 자동 활성화된다.

등록 직후에는 먼저 일반 breakpoint 기반 kprobe를 설치한다. 따라서 특정 지점을 최적화할 수 없어도 probe 자체는 존재한다.

최적화 전에는 jump가 덮을 optimized region 전체가 한 function 안에 있는지, 그 region 안으로 들어오는 jump가 없는지, 각 instruction을 out-of-line으로 실행할 수 있는지 검사한다. Function에 indirect jump나 exception을 일으키는 instruction이 없어야 하며 첫 byte 이외의 위치로 향하는 near jump도 없어야 한다.

검사를 통과하면 detour buffer를 만든다. Buffer에는 breakpoint trap을 흉내 내는 register push, user probe handler를 호출하는 trampoline call, register restore, optimized region에서 복사한 instruction, 원래 실행 경로로 돌아가는 jump가 순서대로 들어간다.

Detour buffer 준비 뒤 probe에 `post_handler`가 있거나 optimized region의 다른 instruction도 probed 상태이거나 probe가 disabled면 최적화를 시작하지 않는다. 이는 임시 상태이므로 조건이 바뀌면 다시 시도한다.

최적화 가능한 kprobe는 optimizing list와 optimizer workqueue에 넣는다. 실제 최적화 전에 적중하면 CPU instruction pointer를 detour buffer의 copied code로 돌려 single-step만이라도 피한다.

Optimizer는 optimized region 실행 중 interrupt된 CPU가 있을 수 있으므로 먼저 `synchronize_rcu()`를 호출한다. 이 보장은 `CONFIG_PREEMPT=n`에서만 성립하므로 이 버전의 optimization도 non-preemptive kernel만 지원한다. 이후 `stop_machine()`과 `text_poke_smp()`로 optimized region을 detour buffer로 향하는 jump로 바꾼다.

Optimized kprobe를 unregister·disable하거나 다른 kprobe가 가로막으면 unoptimization한다. 완료 전이면 list에서 빼고, 완료 뒤라면 `text_poke_smp()`로 첫 byte의 `int3`를 제외한 원래 code를 복원한다.

Jump optimization 상태에서는 `pre_handler`가 `regs->ip`를 바꾸고 1을 반환해도 실행 경로 변경이 무시된다. 실행 경로를 조작하려면 빈 `post_handler`를 지정하거나 `sysctl -w debug.kprobes_optimization=n`으로 optimization을 억제해야 한다.

.. _kprobes_jump_optimization:

How Does Jump Optimization Work?
--------------------------------

If your kernel is built with CONFIG_OPTPROBES=y (currently this flag
is automatically set 'y' on x86/x86-64, non-preemptive kernel) and
the "debug.kprobes_optimization" kernel parameter is set to 1 (see
sysctl(8)), Kprobes tries to reduce probe-hit overhead by using a jump
instruction instead of a breakpoint instruction at each probepoint.

Init a Kprobe
^^^^^^^^^^^^^

When a probe is registered, before attempting this optimization,
Kprobes inserts an ordinary, breakpoint-based kprobe at the specified
address. So, even if it's not possible to optimize this particular
probepoint, there'll be a probe there.

Safety Check
^^^^^^^^^^^^

Before optimizing a probe, Kprobes performs the following safety checks:

- Kprobes verifies that the region that will be replaced by the jump
  instruction (the "optimized region") lies entirely within one function.
  (A jump instruction is multiple bytes, and so may overlay multiple
  instructions.)

- Kprobes analyzes the entire function and verifies that there is no
  jump into the optimized region.  Specifically:

  - the function contains no indirect jump;
  - the function contains no instruction that causes an exception (since
    the fixup code triggered by the exception could jump back into the
    optimized region -- Kprobes checks the exception tables to verify this);
  - there is no near jump to the optimized region (other than to the first
    byte).

- For each instruction in the optimized region, Kprobes verifies that
  the instruction can be executed out of line.

Preparing Detour Buffer
^^^^^^^^^^^^^^^^^^^^^^^

Next, Kprobes prepares a "detour" buffer, which contains the following
instruction sequence:

- code to push the CPU's registers (emulating a breakpoint trap)
- a call to the trampoline code which calls user's probe handlers.
- code to restore registers
- the instructions from the optimized region
- a jump back to the original execution path.

Pre-optimization
^^^^^^^^^^^^^^^^

After preparing the detour buffer, Kprobes verifies that none of the
following situations exist:

- The probe has a post_handler.
- Other instructions in the optimized region are probed.
- The probe is disabled.

In any of the above cases, Kprobes won't start optimizing the probe.
Since these are temporary situations, Kprobes tries to start
optimizing it again if the situation is changed.

If the kprobe can be optimized, Kprobes enqueues the kprobe to an
optimizing list, and kicks the kprobe-optimizer workqueue to optimize
it.  If the to-be-optimized probepoint is hit before being optimized,
Kprobes returns control to the original instruction path by setting
the CPU's instruction pointer to the copied code in the detour buffer
-- thus at least avoiding the single-step.

Optimization
^^^^^^^^^^^^

The Kprobe-optimizer doesn't insert the jump instruction immediately;
rather, it calls synchronize_rcu() for safety first, because it's
possible for a CPU to be interrupted in the middle of executing the
optimized region [3]_.  As you know, synchronize_rcu() can ensure
that all interruptions that were active when synchronize_rcu()
was called are done, but only if CONFIG_PREEMPT=n.  So, this version
of kprobe optimization supports only kernels with CONFIG_PREEMPT=n [4]_.

After that, the Kprobe-optimizer calls stop_machine() to replace
the optimized region with a jump instruction to the detour buffer,
using text_poke_smp().

Unoptimization
^^^^^^^^^^^^^^

When an optimized kprobe is unregistered, disabled, or blocked by
another kprobe, it will be unoptimized.  If this happens before
the optimization is complete, the kprobe is just dequeued from the
optimized list.  If the optimization has been done, the jump is
replaced with the original code (except for an int3 breakpoint in
the first byte) by using text_poke_smp().

.. [3] Please imagine that the 2nd instruction is interrupted and then
   the optimizer replaces the 2nd instruction with the jump *address*
   while the interrupt handler is running. When the interrupt
   returns to original address, there is no valid instruction,
   and it causes an unexpected result.

.. [4] This optimization-safety checking may be replaced with the
   stop-machine method that ksplice uses for supporting a CONFIG_PREEMPT=y
   kernel.

NOTE for geeks:
The jump optimization changes the kprobe's pre_handler behavior.
Without optimization, the pre_handler can change the kernel's execution
path by changing regs->ip and returning 1.  However, when the probe
is optimized, that modification is ignored.  Thus, if you want to
tweak the kernel's execution path, you need to suppress optimization,
using one of the following techniques:

- Specify an empty function for the kprobe's post_handler.

or

- Execute 'sysctl -w debug.kprobes_optimization=n'

Kprobes blacklist

Kprobes는 자기 자신을 제외한 kernel 대부분을 probe할 수 있다. Kprobes 구현 function을 probe하면 recursive trap이나 double fault가 발생하거나 nested probe handler가 호출되지 않을 수 있어 blacklist로 관리한다.

Function을 blacklist에 넣으려면 `linux/kprobes.h`를 include하고 `NOKPROBE_SYMBOL()` macro로 지정한다. 등록 시 주어진 probe address가 blacklist에 있으면 거부된다.

.. _kprobes_blacklist:

Blacklist
---------

Kprobes can probe most of the kernel except itself. This means
that there are some functions where kprobes cannot probe. Probing
(trapping) such functions can cause a recursive trap (e.g. double
fault) or the nested probe handler may never be called.
Kprobes manages such functions as a blacklist.
If you want to add a function into the blacklist, you just need
to (1) include linux/kprobes.h and (2) use NOKPROBE_SYMBOL() macro
to specify a blacklisted function.
Kprobes checks the given probe address against the blacklist and
rejects registering it, if the given address is in the blacklist.

지원 architecture와 kernel 설정

Kprobes와 return probe는 i386, x86_64, ppc64, arm, ppc, mips, s390, parisc, loongarch, riscv에서 구현돼 있다. i386과 x86_64는 jump optimization을 지원하며 sparc64는 아직 return probe를 지원하지 않는다.

Kernel 설정에서 `General architecture-dependent options` 아래 `CONFIG_KPROBES=y`를 선택한다. Instrumentation module을 load·unload하려면 `CONFIG_MODULES=y`와 `CONFIG_MODULE_UNLOAD=y`도 필요하다.

In-kernel address resolution이 `kallsyms_lookup_name()`을 사용하므로 `CONFIG_KALLSYMS`, 필요하면 `CONFIG_KALLSYMS_ALL`도 켠다. Function 중간에 probe를 넣을 때는 `CONFIG_DEBUG_INFO`와 `objdump -d -l vmlinux`로 source-to-object mapping을 확인하면 유용하다.

.. _kprobes_archs_supported:

Architectures Supported
=======================

Kprobes and return probes are implemented on the following
architectures:

- i386 (Supports jump optimization)
- x86_64 (AMD-64, EM64T) (Supports jump optimization)
- ppc64
- sparc64 (Return probes not yet implemented.)
- arm
- ppc
- mips
- s390
- parisc
- loongarch
- riscv

Configuring Kprobes
===================

When configuring the kernel using make menuconfig/xconfig/oldconfig,
ensure that CONFIG_KPROBES is set to "y", look for "Kprobes" under
"General architecture-dependent options".

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".

Also make sure that CONFIG_KALLSYMS and perhaps even CONFIG_KALLSYMS_ALL
are set to "y", since kallsyms_lookup_name() is used by the in-kernel
kprobe address resolution code.

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.

`register_kprobe()`와 handler

Kprobes API는 probe 종류마다 register·unregister function과 배열용 batch function을 제공한다. 실제 예제는 `samples/kprobes/` 아래에서 볼 수 있다.

::

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

`register_kprobe()`는 `kp->addr`에 breakpoint를 설치한다. 적중하면 `kp->pre_handler`, copied instruction single-step 뒤에는 `kp->post_handler`를 호출한다. Handler는 모두 `NULL`일 수 있다. `kp->flags`에 `KPROBE_FLAG_DISABLED`를 설정하면 disabled 상태로 등록되며 `enable_kprobe(kp)` 호출 전에는 handler가 실행되지 않는다.

`struct kprobe.symbol_name`을 사용하면 kernel이 probepoint address를 resolve한다. Symbol 내부 offset을 알면 `offset` field를 사용한다. `symbol_name`과 `addr` 중 하나만 지정해야 하며 둘 다 지정하면 `-EINVAL`이다. CISC architecture에서는 `addr`가 instruction boundary인지 검증하지 않으므로 `offset`을 주의해 사용한다.

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);

`pre_handler`는 breakpoint에 연결된 kprobe와 trap 때 저장한 register를 받는다. 특별히 실행 경로를 바꾸는 경우가 아니면 0을 반환한다.

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);

`post_handler`가 받는 `p`와 `regs`는 pre-handler와 같으며 `flags`는 항상 0으로 보인다. `register_kprobe()`는 성공 시 0, 실패 시 음수 errno를 반환한다.

API Reference
=============

The Kprobes API includes a "register" function and an "unregister"
function for each type of probe. The API also includes "register_*probes"
and "unregister_*probes" functions for (un)registering arrays of probes.
Here are terse, mini-man-page specifications for these functions and
the associated probe handlers that you'll write. See the files in the
samples/kprobes/ sub-directory for examples.

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.  Any or all handlers can be NULL. If kp->flags is set
KPROBE_FLAG_DISABLED, that kp will be registered but disabled, so, its handlers
aren't hit until calling enable_kprobe(kp).

.. note::

   1. With the introduction of the "symbol_name" field to struct kprobe,
      the probepoint address resolution will now be taken care of by the kernel.
      The following will now work::

	kp.symbol_name = "symbol_name";

      (64-bit powerpc intricacies such as function descriptors are handled
      transparently)

   2. Use the "offset" field of struct kprobe if the offset into the symbol
      to install a probepoint is known. This field is used to calculate the
      probepoint.

   3. Specify either the kprobe "symbol_name" OR the "addr". If both are
      specified, kprobe registration will fail with -EINVAL.

   4. With CISC architectures (such as i386 and x86_64), the kprobes code
      does not validate if the kprobe.addr is at an instruction boundary.
      Use "offset" with caution.

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.

`register_kretprobe()`와 return handler

::

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

`register_kretprobe()`는 `rp->kp.addr` function에 return probe를 설치하고 function return 때 `rp->handler`를 호출한다. 등록 전에 `rp->maxactive`를 적절히 설정해야 하며 성공 시 0, 실패 시 음수 errno를 반환한다.

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);

Return handler의 `regs`는 kprobe pre-handler와 같은 저장 register다. `ri`는 `kretprobe_instance`를 가리키며 `ret_addr`, 대응 `kretprobe`인 `rp`, 대상 `task`, instance별 private `data`를 제공한다.

Architecture ABI가 정의한 올바른 register에서 return value를 꺼낼 때는 `regs_return_value(regs)` macro를 사용한다. 현재 handler return value는 무시된다.

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
- data: points to per return-instance private data; see "Kretprobe
	entry-handler" for details.

The regs_return_value(regs) macro provides a simple abstraction to
extract the return value from the appropriate register as defined by
the architecture's ABI.

The handler's return value is currently ignored.

해제·batch·disable·enable API

::

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

`unregister_kprobe()`와 `unregister_kretprobe()`는 등록 이후 언제든 지정 probe를 제거한다. 잘못된 probe, 예를 들어 등록되지 않은 probe를 받으면 그 probe의 `addr` field를 지운다.

::

	#include <linux/kprobes.h>
	int register_kprobes(struct kprobe **kps, int num);
	int register_kretprobes(struct kretprobe **rps, int num);

Batch register API는 pointer 배열의 `num`개 probe를 등록한다. 등록 중 error가 발생하면 실패 probe 이전까지 등록한 모든 probe를 안전하게 해제한 뒤 반환한다. 호출 전 pointer 배열을 할당하고 모든 entry를 채워야 한다.

::

	#include <linux/kprobes.h>
	void unregister_kprobes(struct kprobe **kps, int num);
	void unregister_kretprobes(struct kretprobe **rps, int num);

Batch unregister API는 배열의 probe를 한 번에 제거한다. 배열 안에 잘못된 probe가 있으면 그 entry의 `addr`를 지우지만 나머지 올바른 probe는 정상 해제한다.

::

	#include <linux/kprobes.h>
	int disable_kprobe(struct kprobe *kp);
	int disable_kretprobe(struct kretprobe *rp);

`disable_kprobe()`와 `disable_kretprobe()`는 등록된 probe를 임시 비활성화한다.

::

	#include <linux/kprobes.h>
	int enable_kprobe(struct kprobe *kp);
	int enable_kretprobe(struct kretprobe *rp);

`enable_kprobe()`와 `enable_kretprobe()`는 `disable_*probe()`로 비활성화한 등록 probe를 다시 켠다.

unregister_*probe
------------------

::

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

.. note::

   If the functions find an incorrect probe (ex. an unregistered probe),
   they clear the addr field of the probe.

register_*probes
----------------

::

	#include <linux/kprobes.h>
	int register_kprobes(struct kprobe **kps, int num);
	int register_kretprobes(struct kretprobe **rps, int num);

Registers each of the num probes in the specified array.  If any
error occurs during registration, all probes in the array, up to
the bad probe, are safely unregistered before the register_*probes
function returns.

- kps/rps: an array of pointers to ``*probe`` data structures
- num: the number of the array entries.

.. note::

   You have to allocate(or define) an array of pointers and set all
   of the array entries before using these functions.

unregister_*probes
------------------

::

	#include <linux/kprobes.h>
	void unregister_kprobes(struct kprobe **kps, int num);
	void unregister_kretprobes(struct kretprobe **rps, int num);

Removes each of the num probes in the specified array at once.

.. note::

   If the functions find some incorrect probes (ex. unregistered
   probes) in the specified array, they clear the addr field of those
   incorrect probes. However, other probes in the array are
   unregistered correctly.

disable_*probe
--------------

::

	#include <linux/kprobes.h>
	int disable_kprobe(struct kprobe *kp);
	int disable_kretprobe(struct kretprobe *rp);

Temporarily disables the specified ``*probe``. You can enable it again by using
enable_*probe(). You must specify the probe which has been registered.

enable_*probe
-------------

::

	#include <linux/kprobes.h>
	int enable_kprobe(struct kprobe *kp);
	int enable_kretprobe(struct kretprobe *rp);

Enables ``*probe`` which has been disabled by disable_*probe(). You must specify
the probe which has been registered.

기능과 제약

같은 address에 여러 probe를 설치할 수 있다. `post_handler`가 있는 probepoint는 최적화할 수 없으므로 이미 optimized된 지점에 이런 kprobe를 설치하면 자동으로 unoptimization한다.

대부분의 kernel 위치와 interrupt handler를 probe할 수 있지만 Kprobes 구현 code, `do_page_fault`, `notifier_call_chain` 같은 function은 `-EINVAL`로 거부된다. Inline 가능 function을 probe해도 compiler가 만든 모든 inline instance를 찾아 설치하지는 않는다.

Handler는 kernel data structure나 복귀 때 register에 반영되는 `pt_regs`를 바꿀 수 있어 bug fix나 fault injection에도 쓸 수 있다. 하지만 의도한 fault와 실수를 구분해 주지는 않는다.

Kprobes는 handler 간 충돌을 방지하지 않는다. 예를 들어 `printk()`를 probe한 handler가 다시 `printk()`를 호출해 두 번째 probe를 만나면 그 instance의 두 번째 handler는 실행하지 않고 두 번째 probe의 `nmissed`를 증가시킨다. 서로 다른 CPU에서는 여러 handler가 동시에 실행될 수 있다.

등록·해제 외에는 mutex를 사용하거나 memory를 할당하지 않는다. Handler는 architecture와 optimization 상태에 따라 preemption 또는 interrupt가 disabled된 채 실행되므로 semaphore 획득이나 I/O 대기처럼 CPU를 양보하면 안 된다.

Kretprobe는 return address를 trampoline으로 바꾸므로 stack backtrace와 `__builtin_return_address()`가 실제 return address 대신 trampoline address를 보일 수 있다.

Function 호출 횟수와 return 횟수가 다르거나 current task가 아닌 stack에서 진입·복귀하는 function에 return probe를 걸면 문제가 생길 수 있다. 문제 instance에는 `kretprobe BUG!` 메시지가 출력된다. x86_64의 `__switch_to()`에는 이런 이유로 kprobe와 kretprobe를 지원하지 않고 등록 시 `-EINVAL`을 반환한다.

::

		IA
		|
	[-2][-1][0][1][2][3][4][5][6][7]
		[ins1][ins2][  ins3 ]
		[<-     DCR       ->]
		[<- JTPR ->]

	ins1: 1st Instruction
	ins2: 2nd Instruction
	ins3: 3rd Instruction
	IA:  Insertion Address
	JTPR: Jump Target Prohibition Region
	DCR: Detoured Code Region

x86/x86-64 jump optimization은 5-byte jump가 여러 instruction을 덮으므로 Detoured Code Region(DCR)과 Jump Target Prohibition Region(JTPR) 제약이 있다. DCR instruction은 relocation 가능해야 하고 call instruction을 포함하면 안 된다. JTPR은 jump나 call target이 될 수 없고 DCR은 function 경계를 가로지르면 안 된다. In-kernel instruction decoder가 이를 검사한다.

Kprobes Features and Limitations
================================

Kprobes allows multiple probes at the same address. Also,
a probepoint for which there is a post_handler cannot be optimized.
So if you install a kprobe with a post_handler, at an optimized
probepoint, the probepoint will be unoptimized automatically.

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 or interrupt disabled,
which depends on the architecture and optimization state.  (e.g.,
kretprobe handlers and optimized kprobe handlers run without interrupt
disabled on x86/x86-64).  In any case, your handler should not yield
the CPU (e.g., by attempting to acquire a semaphore, or waiting I/O).

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. In such a case, a line:
kretprobe BUG!: Processing kretprobe d000000000041aa8 @ c00000000004f48c
gets printed. With this information, one will be able to correlate the
exact instance of the kretprobe that caused the problem. 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)
on the x86_64 version of __switch_to(); the registration functions
return -EINVAL.

On x86/x86-64, since the Jump Optimization of Kprobes modifies
instructions widely, there are some limitations to optimization. To
explain it, we introduce some terminology. Imagine a 3-instruction
sequence consisting of a two 2-byte instructions and one 3-byte
instruction.

::

		IA
		|
	[-2][-1][0][1][2][3][4][5][6][7]
		[ins1][ins2][  ins3 ]
		[<-     DCR       ->]
		[<- JTPR ->]

	ins1: 1st Instruction
	ins2: 2nd Instruction
	ins3: 3rd Instruction
	IA:  Insertion Address
	JTPR: Jump Target Prohibition Region
	DCR: Detoured Code Region

The instructions in DCR are copied to the out-of-line buffer
of the kprobe, because the bytes in DCR are replaced by
a 5-byte jump instruction. So there are several limitations.

a) The instructions in DCR must be relocatable.
b) The instructions in DCR must not include a call instruction.
c) JTPR must not be targeted by any jump or call instruction.
d) DCR must not straddle the border between functions.

Anyway, these limitations are checked by the in-kernel instruction
decoder, so you don't need to worry about that.

Probe overhead

2005년 일반 CPU 기준 kprobe hit 처리 시간은 0.5~1.0 microseconds이며 단순 handler benchmark는 architecture에 따라 초당 100만~200만 hit를 기록했다. Return probe hit는 일반 kprobe보다 보통 50~75% 더 오래 걸린다. Function에 return probe가 이미 있을 때 entry kprobe를 추가하는 overhead는 사실상 없다.

Here are sample overhead figures (in usec) for different architectures::

  k = kprobe; r = return probe; kr = kprobe + return probe
  on same function

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

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

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

Optimized kprobe hit는 보통 0.07~0.1 microseconds다. 원문 benchmark는 unoptimized, boosted, optimized kprobe와 kretprobe를 구분한다.

Typically, an optimized kprobe hit takes 0.07 to 0.1 microseconds to
process. Here are sample overhead figures (in usec) for x86 architectures::

  k = unoptimized kprobe, b = boosted (single-step skipped), o = optimized kprobe,
  r = unoptimized kretprobe, rb = boosted kretprobe, ro = optimized kretprobe.

  i386: Intel(R) Xeon(R) E5410, 2.33GHz, 4656.90 bogomips
  k = 0.80 usec; b = 0.33; o = 0.05; r = 1.10; rb = 0.61; ro = 0.33

  x86-64: Intel(R) Xeon(R) E5410, 2.33GHz, 4656.90 bogomips
  k = 0.99 usec; b = 0.43; o = 0.06; r = 1.24; rb = 0.68; ro = 0.30

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 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; r = return probe; kr = kprobe + return probe
  on same function

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

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

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

Optimized Probe Overhead
------------------------

Typically, an optimized kprobe hit takes 0.07 to 0.1 microseconds to
process. Here are sample overhead figures (in usec) for x86 architectures::

  k = unoptimized kprobe, b = boosted (single-step skipped), o = optimized kprobe,
  r = unoptimized kretprobe, rb = boosted kretprobe, ro = optimized kretprobe.

  i386: Intel(R) Xeon(R) E5410, 2.33GHz, 4656.90 bogomips
  k = 0.80 usec; b = 0.33; o = 0.05; r = 1.10; rb = 0.61; ro = 0.33

  x86-64: Intel(R) Xeon(R) E5410, 2.33GHz, 4656.90 bogomips
  k = 0.99 usec; b = 0.43; o = 0.06; r = 1.24; rb = 0.68; ro = 0.30

TODO, 예제와 deprecated 기능

TODO 목록에는 SystemTap의 간소화 interface 활용, sparc64 return probe, 추가 architecture, user-space probe, data reference에 반응하는 watchpoint probe 지원이 있다.

일반 kprobe module 예제는 `samples/kprobes/kprobe_example.c`, kretprobe 예제는 `samples/kprobes/kretprobe_example.c`를 참조한다.

Jprobes는 deprecated됐다. 의존하는 사용자는 다른 tracing 기능으로 이동하거나 구형 kernel을 사용해야 한다.

첫 대안은 argument가 포함된 target function을 정적으로 정의한 low-overhead trace event로 추적하는 것이다. Event가 꺼져 있으면 눈에 보이는 overhead가 거의 없으며 ftrace나 다른 tracing tool로 사용할 수 있다.

두 번째 대안은 `perf-probe`와 ftrace dynamic kprobe event다. `CONFIG_DEBUG_INFO=y` kernel에서는 `perf-probe`가 local variable과 argument에 할당된 register·stack 위치를 찾아 새 event를 구성할 수 있다. `Documentation/trace/kprobetrace.rst`, `Documentation/trace/events.rst`, `tools/perf/Documentation/perf-probe.txt`를 참조한다.

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).

Kprobes Example
===============

See samples/kprobes/kprobe_example.c

Kretprobes Example
==================

See samples/kprobes/kretprobe_example.c

Deprecated Features
===================

Jprobes is now a deprecated feature. People who are depending on it should
migrate to other tracing features or use older kernels. Please consider to
migrate your tool to one of the following options:

- Use trace-event to trace target function with arguments.

  trace-event is a low-overhead (and almost no visible overhead if it
  is off) statically defined event interface. You can define new events
  and trace it via ftrace or any other tracing tools.

  See the following urls:

    - https://lwn.net/Articles/379903/
    - https://lwn.net/Articles/381064/
    - https://lwn.net/Articles/383362/

- Use ftrace dynamic events (kprobe event) with perf-probe.

  If you build your kernel with debug info (CONFIG_DEBUG_INFO=y), you can
  find which register/stack is assigned to which local variable or arguments
  by using perf-probe and set up new event to trace it.

  See following documents:

  - Documentation/trace/kprobetrace.rst
  - Documentation/trace/events.rst
  - tools/perf/Documentation/perf-probe.txt

Kprobes debugfs interface

최근 kernel에서는 debugfs가 mount돼 있으면 `/sys/kernel/debug/kprobes/` 아래에서 등록된 kprobe 목록을 볼 수 있다.

/sys/kernel/debug/kprobes/list: Lists all registered probes on the system::

	c015d71a  k  vfs_read+0x0
	c03dedc5  r  tcp_v4_rcv+0x0

`list`의 첫 column은 probe가 삽입된 kernel address, 둘째는 `k` 또는 `r`인 probe 종류, 셋째는 `symbol+offset`이다. Module function이면 module 이름도 표시한다.

뒤 status column의 `[GONE]`은 module unload 등으로 virtual address가 더 이상 유효하지 않음을, `[DISABLED]`는 임시 비활성화를, `[OPTIMIZED]`는 jump optimization을, `[FTRACE]`는 ftrace 기반 probe를 뜻한다.

`/sys/kernel/debug/kprobes/enabled`는 등록된 모든 kprobe를 전역으로 강제 ON/OFF한다. 0을 쓰면 모두 disarm하고 1을 쓰면 다시 arm하지만 각 probe의 disabled state는 바꾸지 않는다. 따라서 `[DISABLED]` probe는 전역 ON 뒤에도 활성화되지 않는다.

The kprobes debugfs interface
=============================


With recent kernels (> 2.6.20) the list of registered kprobes is visible
under the /sys/kernel/debug/kprobes/ directory (assuming debugfs is mounted at //sys/kernel/debug).

/sys/kernel/debug/kprobes/list: Lists all registered probes on the system::

	c015d71a  k  vfs_read+0x0
	c03dedc5  r  tcp_v4_rcv+0x0

The first column provides the kernel address where the probe is inserted.
The second column identifies the type of probe (k - kprobe and r - kretprobe)
while the third column specifies the symbol+offset of the probe.
If the probed function belongs to a module, the module name is also
specified. Following columns show probe status. If the probe is on
a virtual address that is no longer valid (module init sections, module
virtual addresses that correspond to modules that've been unloaded),
such probes are marked with [GONE]. If the probe is temporarily disabled,
such probes are marked with [DISABLED]. If the probe is optimized, it is
marked with [OPTIMIZED]. If the probe is ftrace-based, it is marked with
[FTRACE].

/sys/kernel/debug/kprobes/enabled: Turn kprobes ON/OFF forcibly.

Provides a knob to globally and forcibly turn registered kprobes ON or OFF.
By default, all kprobes are enabled. By echoing "0" to this file, all
registered probes will be disarmed, till such time a "1" is echoed to this
file. Note that this knob just disarms and arms all kprobes and doesn't
change each probe's disabling state. This means that disabled kprobes (marked
[DISABLED]) will be not enabled if you turn ON all kprobes by this knob.

Optimization sysctl과 참고 자료

`CONFIG_OPTPROBES=y`이면 `/proc/sys/debug/kprobes-optimization`이 나타나며 jump optimization을 전역 ON/OFF한다. 기본값은 ON이다.

파일에 0을 쓰거나 sysctl로 `debug.kprobes_optimization=0`을 설정하면 모든 optimized probe가 unoptimization되고 이후 등록되는 probe도 최적화하지 않는다. 이 knob는 단순 허용 여부가 아니라 현재 optimized state를 실제로 바꾸므로 `[OPTIMIZED]` tag가 제거되며 다시 켜면 재최적화된다.

추가 정보는 원문에 명시된 LWN Kprobes 문서와 OLS 2006 PDF를 참조한다.

The kprobes sysctl interface
============================

/proc/sys/debug/kprobes-optimization: Turn kprobes optimization ON/OFF.

When CONFIG_OPTPROBES=y, this sysctl interface appears and it provides
a knob to globally and forcibly turn jump optimization (see section
:ref:`kprobes_jump_optimization`) ON or OFF. By default, jump optimization
is allowed (ON). If you echo "0" to this file or set
"debug.kprobes_optimization" to 0 via sysctl, all optimized probes will be
unoptimized, and any new probes registered after that will not be optimized.

Note that this knob *changes* the optimized state. This means that optimized
probes (marked [OPTIMIZED]) will be unoptimized ([OPTIMIZED] tag will be
removed). If the knob is turned on, they will be optimized again.

References
==========

For additional information on Kprobes, refer to the following URLs:

- https://lwn.net/Articles/132196/
- https://www.kernel.org/doc/ols/2006/ols2006v2-pages-109-124.pdf