How long time does it need to detect the heartbleed vul?

[Mambaboy]

Hi, I have done the experiment as the tutorial, but only about 78 paths detected after 24 hours fuzzing, neither does the heartbleed vulnerability. The process is as follow:

1. git clone https://github.com/openssl/openssl.git
2. git checkout Checkout at tag OpenSSL_1_0_1f
3. export CC=/afl/afl-clang-fast export CXX=afl/afl-clang-fast++
4. in openssl dir, ./config && AFL_USE_ASAN=1 make
5. build the handshake binary
   AFL_USE_ASAN=1 /afl/afl-clang-fast++ -g handshake.cc openssl/libssl.a openssl/libcrypto.a -o handshake -I openssl/include -ldl
6. set "AAAAAAAAAA" as the seed
7. run the AFL
   /afl/afl-fuzz -i /seed -o /out -m none -- //handshake

Regards, xiaosatianyu

[mykter]

It took a minute or two when I ran it. Will see if I can reproduce.

[mykter]

That's using afl-2.52 and clang-4.0. Similar to what I remember when I first created it.

I just tried again with your seed, and this time it found it in <1minute, 18 paths. So doesn't look like a fluke, and certainly not one that would survive 24hrs fuzzing.

Standard afl debugging required then:

• what happens if you run the handshake under afl-showmap instead of afl-fuzz - any errors? e.g. ~/local/afl-2.52b/afl-showmap -o map -m none ./handshake < in/1
• when you compiled openssl, did you see all the notifications from afl-clang-fast confirming ASAN mode was used like [+] Instrumented 17 locations (ASAN/MSAN mode, ratio 100%).
• are there any warning indicators in the afl-fuzz status screen? Anything yellow or red?
• What did you use for your harness? The example from ANSWERS.md or something else?

(I noticed your steps didn't specify "./config.sh -d" - it's worth including the -d so you've got debug symbols when you do find a crash. Update: turns out that doesn't necessarily help, and you need to pass the log through asan_symbolize.)

[Mambaboy]

Thanks for your reply. I review the process and find there is a change in my code. When compiling the handshake.cc modified according to ANSWERS.md, it fails and outputs an error that

SSL routines:SSL_CTX_use_certificate:ee key too small:ssl/ssl_rsa.c:310

So I add a more line code:

SSL_CTX_set_security_level(sctx, 0);

Then it compiles successfully. After the modification, the experiment is done entirely according to the tutorial. However it still does not detect the crash. Is the added code wrong? Or it would impede the detection of heartbleed bug?

My handshake.cc is as follow:

// Copyright 2016 Google Inc. All Rights Reserved.
// Licensed under the Apache License, Version 2.0 (the "License");
#include <openssl/ssl.h>
#include <openssl/err.h>
#include <assert.h>
#include <stdint.h>
#include <stddef.h>
#include <unistd.h>

#ifndef CERT_PATH
# define CERT_PATH
#endif

SSL_CTX *Init() {
  SSL_library_init();
  SSL_load_error_strings();
  ERR_load_BIO_strings();
  OpenSSL_add_all_algorithms();
  SSL_CTX *sctx;
  assert (sctx = SSL_CTX_new(TLSv1_method()));
  SSL_CTX_set_security_level(sctx, 0);
  /* These two file were created with this command:
      openssl req -x509 -newkey rsa:512 -keyout server.key \
     -out server.pem -days 9999 -nodes -subj /CN=a/
  */
        if (!SSL_CTX_use_certificate_file(sctx, "./server.pem",  SSL_FILETYPE_PEM)){
            ERR_print_errors_fp(stderr);
            exit(1);        
        }

  assert(SSL_CTX_use_certificate_file(sctx, "./server.pem",
                                      SSL_FILETYPE_PEM));
  assert(SSL_CTX_use_PrivateKey_file(sctx, "./server.key",
                                     SSL_FILETYPE_PEM));
  return sctx;
}

int main() {
  static SSL_CTX *sctx = Init();
  SSL *server = SSL_new(sctx);
  BIO *sinbio = BIO_new(BIO_s_mem());
  BIO *soutbio = BIO_new(BIO_s_mem());
  SSL_set_bio(server, sinbio, soutbio);
  SSL_set_accept_state(server);

  /* TODO: To spoof one end of the handshake, we need to write data to sinbio
   * here */
    #ifdef __AFL_HAVE_MANUAL_CONTROL
        __AFL_INIT();
    #endif

    uint8_t data[100] = {0};
    size_t size = read(STDIN_FILENO, data, 100);
    if (size == -1) {
        printf("Failed to read from stdin\n");
        return(-1);
    }
  BIO_write(sinbio, data, size);

  SSL_do_handshake(server);
  SSL_free(server);
  return 0;
}

[mykter]

I didn't have to make a modification when I tested it earlier - are you definitely on the right version of OpenSSL? That would explain both the harness needing changes and heartbleed not being there.

[mykter]

I'm going to close this, let me know if you're still having issues.