blob: 8888cbeac72ea10e2a7aab4c8ae3429f86878d45 [file] [log] [blame]
// -*- C++ -*-
//===------------------------- fuzzing.cpp -------------------------------===//
//
// The LLVM Compiler Infrastructure
//
// This file is dual licensed under the MIT and the University of Illinois Open
// Source Licenses. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
// A set of routines to use when fuzzing the algorithms in libc++
// Each one tests a single algorithm.
//
// They all have the form of:
// int `algorithm`(const uint8_t *data, size_t size);
//
// They perform the operation, and then check to see if the results are correct.
// If so, they return zero, and non-zero otherwise.
//
// For example, sort calls std::sort, then checks two things:
// (1) The resulting vector is sorted
// (2) The resulting vector contains the same elements as the original data.
#include "fuzzing.h"
#include <vector>
#include <algorithm>
#include <functional>
#include <regex>
#include <cassert>
#include <iostream>
// If we had C++14, we could use the four iterator version of is_permutation and equal
namespace fuzzing {
// This is a struct we can use to test the stable_XXX algorithms.
// perform the operation on the key, then check the order of the payload.
struct stable_test {
uint8_t key;
size_t payload;
stable_test(uint8_t k) : key(k), payload(0) {}
stable_test(uint8_t k, size_t p) : key(k), payload(p) {}
};
void swap(stable_test &lhs, stable_test &rhs)
{
using std::swap;
swap(lhs.key, rhs.key);
swap(lhs.payload, rhs.payload);
}
struct key_less
{
bool operator () (const stable_test &lhs, const stable_test &rhs) const
{
return lhs.key < rhs.key;
}
};
struct payload_less
{
bool operator () (const stable_test &lhs, const stable_test &rhs) const
{
return lhs.payload < rhs.payload;
}
};
struct total_less
{
bool operator () (const stable_test &lhs, const stable_test &rhs) const
{
return lhs.key == rhs.key ? lhs.payload < rhs.payload : lhs.key < rhs.key;
}
};
bool operator==(const stable_test &lhs, const stable_test &rhs)
{
return lhs.key == rhs.key && lhs.payload == rhs.payload;
}
template<typename T>
struct is_even
{
bool operator () (const T &t) const
{
return t % 2 == 0;
}
};
template<>
struct is_even<stable_test>
{
bool operator () (const stable_test &t) const
{
return t.key % 2 == 0;
}
};
typedef std::vector<uint8_t> Vec;
typedef std::vector<stable_test> StableVec;
typedef StableVec::const_iterator SVIter;
// Cheap version of is_permutation
// Builds a set of buckets for each of the key values.
// Sums all the payloads.
// Not 100% perfect, but _way_ faster
bool is_permutation(SVIter first1, SVIter last1, SVIter first2)
{
size_t xBuckets[256] = {0};
size_t xPayloads[256] = {0};
size_t yBuckets[256] = {0};
size_t yPayloads[256] = {0};
for (; first1 != last1; ++first1, ++first2)
{
xBuckets [first1->key]++;
xPayloads[first1->key] += first1->payload;
yBuckets [first2->key]++;
yPayloads[first2->key] += first2->payload;
}
for (size_t i = 0; i < 256; ++i)
{
if (xBuckets[i] != yBuckets[i])
return false;
if (xPayloads[i] != yPayloads[i])
return false;
}
return true;
}
template <typename Iter1, typename Iter2>
bool is_permutation(Iter1 first1, Iter1 last1, Iter2 first2)
{
static_assert((std::is_same<typename std::iterator_traits<Iter1>::value_type, uint8_t>::value), "");
static_assert((std::is_same<typename std::iterator_traits<Iter2>::value_type, uint8_t>::value), "");
size_t xBuckets[256] = {0};
size_t yBuckets[256] = {0};
for (; first1 != last1; ++first1, ++first2)
{
xBuckets [*first1]++;
yBuckets [*first2]++;
}
for (size_t i = 0; i < 256; ++i)
if (xBuckets[i] != yBuckets[i])
return false;
return true;
}
// == sort ==
int sort(const uint8_t *data, size_t size)
{
Vec working(data, data + size);
std::sort(working.begin(), working.end());
if (!std::is_sorted(working.begin(), working.end())) return 1;
if (!fuzzing::is_permutation(data, data + size, working.cbegin())) return 99;
return 0;
}
// == stable_sort ==
int stable_sort(const uint8_t *data, size_t size)
{
StableVec input;
for (size_t i = 0; i < size; ++i)
input.push_back(stable_test(data[i], i));
StableVec working = input;
std::stable_sort(working.begin(), working.end(), key_less());
if (!std::is_sorted(working.begin(), working.end(), key_less())) return 1;
auto iter = working.begin();
while (iter != working.end())
{
auto range = std::equal_range(iter, working.end(), *iter, key_less());
if (!std::is_sorted(range.first, range.second, total_less())) return 2;
iter = range.second;
}
if (!fuzzing::is_permutation(input.cbegin(), input.cend(), working.cbegin())) return 99;
return 0;
}
// == partition ==
int partition(const uint8_t *data, size_t size)
{
Vec working(data, data + size);
auto iter = std::partition(working.begin(), working.end(), is_even<uint8_t>());
if (!std::all_of (working.begin(), iter, is_even<uint8_t>())) return 1;
if (!std::none_of(iter, working.end(), is_even<uint8_t>())) return 2;
if (!fuzzing::is_permutation(data, data + size, working.cbegin())) return 99;
return 0;
}
// == partition_copy ==
int partition_copy(const uint8_t *data, size_t size)
{
Vec v1, v2;
auto iter = std::partition_copy(data, data + size,
std::back_inserter<Vec>(v1), std::back_inserter<Vec>(v2),
is_even<uint8_t>());
// The two vectors should add up to the original size
if (v1.size() + v2.size() != size) return 1;
// All of the even values should be in the first vector, and none in the second
if (!std::all_of (v1.begin(), v1.end(), is_even<uint8_t>())) return 2;
if (!std::none_of(v2.begin(), v2.end(), is_even<uint8_t>())) return 3;
// Every value in both vectors has to be in the original
for (auto v: v1)
if (std::find(data, data + size, v) == data + size) return 4;
for (auto v: v2)
if (std::find(data, data + size, v) == data + size) return 5;
return 0;
}
// == stable_partition ==
int stable_partition (const uint8_t *data, size_t size)
{
StableVec input;
for (size_t i = 0; i < size; ++i)
input.push_back(stable_test(data[i], i));
StableVec working = input;
auto iter = std::stable_partition(working.begin(), working.end(), is_even<stable_test>());
if (!std::all_of (working.begin(), iter, is_even<stable_test>())) return 1;
if (!std::none_of(iter, working.end(), is_even<stable_test>())) return 2;
if (!std::is_sorted(working.begin(), iter, payload_less())) return 3;
if (!std::is_sorted(iter, working.end(), payload_less())) return 4;
if (!fuzzing::is_permutation(input.cbegin(), input.cend(), working.cbegin())) return 99;
return 0;
}
// == nth_element ==
// use the first element as a position into the data
int nth_element (const uint8_t *data, size_t size)
{
if (size <= 1) return 0;
const size_t partition_point = data[0] % size;
Vec working(data + 1, data + size);
const auto partition_iter = working.begin() + partition_point;
std::nth_element(working.begin(), partition_iter, working.end());
// nth may be the end iterator, in this case nth_element has no effect.
if (partition_iter == working.end())
{
if (!std::equal(data + 1, data + size, working.begin())) return 98;
}
else
{
const uint8_t nth = *partition_iter;
if (!std::all_of(working.begin(), partition_iter, [=](uint8_t v) { return v <= nth; }))
return 1;
if (!std::all_of(partition_iter, working.end(), [=](uint8_t v) { return v >= nth; }))
return 2;
if (!fuzzing::is_permutation(data + 1, data + size, working.cbegin())) return 99;
}
return 0;
}
// == partial_sort ==
// use the first element as a position into the data
int partial_sort (const uint8_t *data, size_t size)
{
if (size <= 1) return 0;
const size_t sort_point = data[0] % size;
Vec working(data + 1, data + size);
const auto sort_iter = working.begin() + sort_point;
std::partial_sort(working.begin(), sort_iter, working.end());
if (sort_iter != working.end())
{
const uint8_t nth = *std::min_element(sort_iter, working.end());
if (!std::all_of(working.begin(), sort_iter, [=](uint8_t v) { return v <= nth; }))
return 1;
if (!std::all_of(sort_iter, working.end(), [=](uint8_t v) { return v >= nth; }))
return 2;
}
if (!std::is_sorted(working.begin(), sort_iter)) return 3;
if (!fuzzing::is_permutation(data + 1, data + size, working.cbegin())) return 99;
return 0;
}
// == partial_sort_copy ==
// use the first element as a count
int partial_sort_copy (const uint8_t *data, size_t size)
{
if (size <= 1) return 0;
const size_t num_results = data[0] % size;
Vec results(num_results);
(void) std::partial_sort_copy(data + 1, data + size, results.begin(), results.end());
// The results have to be sorted
if (!std::is_sorted(results.begin(), results.end())) return 1;
// All the values in results have to be in the original data
for (auto v: results)
if (std::find(data + 1, data + size, v) == data + size) return 2;
// The things in results have to be the smallest N in the original data
Vec sorted(data + 1, data + size);
std::sort(sorted.begin(), sorted.end());
if (!std::equal(results.begin(), results.end(), sorted.begin())) return 3;
return 0;
}
// The second sequence has been "uniqued"
template <typename Iter1, typename Iter2>
static bool compare_unique(Iter1 first1, Iter1 last1, Iter2 first2, Iter2 last2)
{
assert(first1 != last1 && first2 != last2);
if (*first1 != *first2) return false;
uint8_t last_value = *first1;
++first1; ++first2;
while(first1 != last1 && first2 != last2)
{
// Skip over dups in the first sequence
while (*first1 == last_value)
if (++first1 == last1) return false;
if (*first1 != *first2) return false;
last_value = *first1;
++first1; ++first2;
}
// Still stuff left in the 'uniqued' sequence - oops
if (first1 == last1 && first2 != last2) return false;
// Still stuff left in the original sequence - better be all the same
while (first1 != last1)
{
if (*first1 != last_value) return false;
++first1;
}
return true;
}
// == unique ==
int unique (const uint8_t *data, size_t size)
{
Vec working(data, data + size);
std::sort(working.begin(), working.end());
Vec results = working;
Vec::iterator new_end = std::unique(results.begin(), results.end());
Vec::iterator it; // scratch iterator
// Check the size of the unique'd sequence.
// it should only be zero if the input sequence was empty.
if (results.begin() == new_end)
return working.size() == 0 ? 0 : 1;
// 'results' is sorted
if (!std::is_sorted(results.begin(), new_end)) return 2;
// All the elements in 'results' must be different
it = results.begin();
uint8_t prev_value = *it++;
for (; it != new_end; ++it)
{
if (*it == prev_value) return 3;
prev_value = *it;
}
// Every element in 'results' must be in 'working'
for (it = results.begin(); it != new_end; ++it)
if (std::find(working.begin(), working.end(), *it) == working.end())
return 4;
// Every element in 'working' must be in 'results'
for (auto v : working)
if (std::find(results.begin(), new_end, v) == new_end)
return 5;
return 0;
}
// == unique_copy ==
int unique_copy (const uint8_t *data, size_t size)
{
Vec working(data, data + size);
std::sort(working.begin(), working.end());
Vec results;
(void) std::unique_copy(working.begin(), working.end(),
std::back_inserter<Vec>(results));
Vec::iterator it; // scratch iterator
// Check the size of the unique'd sequence.
// it should only be zero if the input sequence was empty.
if (results.size() == 0)
return working.size() == 0 ? 0 : 1;
// 'results' is sorted
if (!std::is_sorted(results.begin(), results.end())) return 2;
// All the elements in 'results' must be different
it = results.begin();
uint8_t prev_value = *it++;
for (; it != results.end(); ++it)
{
if (*it == prev_value) return 3;
prev_value = *it;
}
// Every element in 'results' must be in 'working'
for (auto v : results)
if (std::find(working.begin(), working.end(), v) == working.end())
return 4;
// Every element in 'working' must be in 'results'
for (auto v : working)
if (std::find(results.begin(), results.end(), v) == results.end())
return 5;
return 0;
}
// -- regex fuzzers
static int regex_helper(const uint8_t *data, size_t size, std::regex::flag_type flag)
{
if (size > 0)
{
try
{
std::string s((const char *)data, size);
std::regex re(s, flag);
return std::regex_match(s, re) ? 1 : 0;
}
catch (std::regex_error &ex) {}
}
return 0;
}
int regex_ECMAScript (const uint8_t *data, size_t size)
{
(void) regex_helper(data, size, std::regex_constants::ECMAScript);
return 0;
}
int regex_POSIX (const uint8_t *data, size_t size)
{
(void) regex_helper(data, size, std::regex_constants::basic);
return 0;
}
int regex_extended (const uint8_t *data, size_t size)
{
(void) regex_helper(data, size, std::regex_constants::extended);
return 0;
}
int regex_awk (const uint8_t *data, size_t size)
{
(void) regex_helper(data, size, std::regex_constants::awk);
return 0;
}
int regex_grep (const uint8_t *data, size_t size)
{
(void) regex_helper(data, size, std::regex_constants::grep);
return 0;
}
int regex_egrep (const uint8_t *data, size_t size)
{
(void) regex_helper(data, size, std::regex_constants::egrep);
return 0;
}
// -- heap fuzzers
int make_heap (const uint8_t *data, size_t size)
{
Vec working(data, data + size);
std::make_heap(working.begin(), working.end());
if (!std::is_heap(working.begin(), working.end())) return 1;
if (!fuzzing::is_permutation(data, data + size, working.cbegin())) return 99;
return 0;
}
int push_heap (const uint8_t *data, size_t size)
{
if (size < 2) return 0;
// Make a heap from the first half of the data
Vec working(data, data + size);
auto iter = working.begin() + (size / 2);
std::make_heap(working.begin(), iter);
if (!std::is_heap(working.begin(), iter)) return 1;
// Now push the rest onto the heap, one at a time
++iter;
for (; iter != working.end(); ++iter) {
std::push_heap(working.begin(), iter);
if (!std::is_heap(working.begin(), iter)) return 2;
}
if (!fuzzing::is_permutation(data, data + size, working.cbegin())) return 99;
return 0;
}
int pop_heap (const uint8_t *data, size_t size)
{
if (size < 2) return 0;
Vec working(data, data + size);
std::make_heap(working.begin(), working.end());
// Pop things off, one at a time
auto iter = --working.end();
while (iter != working.begin()) {
std::pop_heap(working.begin(), iter);
if (!std::is_heap(working.begin(), --iter)) return 2;
}
return 0;
}
// -- search fuzzers
int search (const uint8_t *data, size_t size)
{
if (size < 2) return 0;
const size_t pat_size = data[0] * (size - 1) / std::numeric_limits<uint8_t>::max();
assert(pat_size <= size - 1);
const uint8_t *pat_begin = data + 1;
const uint8_t *pat_end = pat_begin + pat_size;
const uint8_t *data_end = data + size;
assert(pat_end <= data_end);
// std::cerr << "data[0] = " << size_t(data[0]) << " ";
// std::cerr << "Pattern size = " << pat_size << "; corpus is " << size - 1 << std::endl;
auto it = std::search(pat_end, data_end, pat_begin, pat_end);
if (it != data_end) // not found
if (!std::equal(pat_begin, pat_end, it))
return 1;
return 0;
}
template <typename S>
static int search_helper (const uint8_t *data, size_t size)
{
if (size < 2) return 0;
const size_t pat_size = data[0] * (size - 1) / std::numeric_limits<uint8_t>::max();
const uint8_t *pat_begin = data + 1;
const uint8_t *pat_end = pat_begin + pat_size;
const uint8_t *data_end = data + size;
auto it = std::search(pat_end, data_end, S(pat_begin, pat_end));
if (it != data_end) // not found
if (!std::equal(pat_begin, pat_end, it))
return 1;
return 0;
}
// These are still in std::experimental
// int search_boyer_moore (const uint8_t *data, size_t size)
// {
// return search_helper<std::boyer_moore_searcher<const uint8_t *>>(data, size);
// }
//
// int search_boyer_moore_horspool (const uint8_t *data, size_t size)
// {
// return search_helper<std::boyer_moore_horspool_searcher<const uint8_t *>>(data, size);
// }
// -- set operation fuzzers
template <typename S>
static void set_helper (const uint8_t *data, size_t size, Vec &v1, Vec &v2)
{
assert(size > 1);
const size_t pat_size = data[0] * (size - 1) / std::numeric_limits<uint8_t>::max();
const uint8_t *pat_begin = data + 1;
const uint8_t *pat_end = pat_begin + pat_size;
const uint8_t *data_end = data + size;
v1.assign(pat_begin, pat_end);
v2.assign(pat_end, data_end);
std::sort(v1.begin(), v1.end());
std::sort(v2.begin(), v2.end());
}
} // namespace fuzzing