cgyurgyik | e14a7ff | 2020-07-29 16:29:19 -0400 | [diff] [blame] | 1 | //===-- strstr_fuzz.cpp ---------------------------------------------------===// |
| 2 | // |
| 3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
| 4 | // See https://llvm.org/LICENSE.txt for license information. |
| 5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
| 6 | // |
| 7 | //===----------------------------------------------------------------------===// |
| 8 | /// |
| 9 | /// Fuzzing test for llvm-libc strstr implementation. |
| 10 | /// |
| 11 | //===----------------------------------------------------------------------===// |
| 12 | |
| 13 | #include "src/string/strlen.h" |
| 14 | #include "src/string/strstr.h" |
| 15 | #include <stddef.h> |
| 16 | #include <stdint.h> |
| 17 | |
| 18 | // Simple loop to compare two strings up to a size n. |
| 19 | static int simple_memcmp(const char *left, const char *right, size_t n) { |
| 20 | for (; n && *left == *right; ++left, ++right, --n) |
| 21 | ; |
| 22 | return n ? *left - *right : 0; |
| 23 | } |
| 24 | |
| 25 | // The general structure is to take the value of the first byte, set size1 to |
| 26 | // that value, and add the null terminator. size2 will then contain the rest of |
| 27 | // the bytes in data. |
| 28 | // For example, with inputs (data={2, 6, 4, 8, 0}, size=5): |
| 29 | // size1: data[0] = 2 |
| 30 | // data1: {2, 6} + '\0' = {2, 6, '\0'} |
| 31 | // size2: size - size1 = 3 |
| 32 | // data2: {4, 8, '\0'} |
| 33 | extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) { |
| 34 | // Verify the size is at least 1 and the data is null terminated. |
| 35 | if (!size || data[size - 1] != '\0') |
| 36 | return 0; |
| 37 | const size_t size1 = (data[0] <= size ? data[0] : size); |
| 38 | // The first size will always be at least 1 since |
| 39 | // we need to append the null terminator. The second size |
| 40 | // needs to be checked since it must also contain the null |
| 41 | // terminator. |
| 42 | if (size - size1 == 0) |
| 43 | return 0; |
| 44 | |
| 45 | // Copy the data into a new container. |
| 46 | uint8_t *container = new uint8_t[size1 + 1]; |
| 47 | if (!container) |
| 48 | __builtin_trap(); |
| 49 | |
| 50 | size_t i; |
| 51 | for (i = 0; i < size1; ++i) |
| 52 | container[i] = data[i]; |
| 53 | container[size1] = '\0'; // Add null terminator to container. |
| 54 | |
| 55 | const char *needle = reinterpret_cast<const char *>(container); |
| 56 | const char *haystack = reinterpret_cast<const char *>(data + i); |
| 57 | const char *result = __llvm_libc::strstr(haystack, needle); |
| 58 | |
| 59 | // A null terminator may exist earlier in each, so this needs to be recorded. |
| 60 | const size_t haystack_size = __llvm_libc::strlen(haystack); |
| 61 | const size_t needle_size = __llvm_libc::strlen(needle); |
| 62 | |
| 63 | if (result) { |
| 64 | // The needle is in the haystack. |
| 65 | // 1. Verify that the result matches the needle. |
| 66 | if (simple_memcmp(needle, result, needle_size) != 0) |
| 67 | __builtin_trap(); |
| 68 | |
| 69 | const char *haystack_ptr = haystack; |
| 70 | // 2. Verify that the result is the first occurrence of the needle. |
| 71 | for (; haystack_ptr != result; ++haystack_ptr) { |
| 72 | if (simple_memcmp(needle, haystack_ptr, needle_size) == 0) |
| 73 | __builtin_trap(); // There was an earlier occurrence of the needle. |
| 74 | } |
| 75 | } else { |
| 76 | // No result was found. Verify that the needle doesn't exist within the |
| 77 | // haystack. |
| 78 | for (size_t i = 0; i + needle_size < haystack_size; ++i) { |
| 79 | if (simple_memcmp(needle, haystack + i, needle_size) == 0) |
| 80 | __builtin_trap(); // There was an earlier occurrence of the needle. |
| 81 | } |
| 82 | } |
| 83 | delete[] container; |
| 84 | return 0; |
| 85 | } |