merge(3p/immer): Subtree merge at 'ad3e3556d' as 'third_party/immer' r/1299

Change-Id: I9636a41ad44b4218293833fd3e9456d9b07c731b

1 files changed, 364 insertions, 0 deletions

diff --git a/third_party/immer/test/flex_vector/fuzzed-4.cpp b/third_party/immer/test/flex_vector/fuzzed-4.cpp
new file mode 100644
index 000000000000..d9ecdd7fba7f
--- /dev/null
+++ b/third_party/immer/test/flex_vector/fuzzed-4.cpp
@@ -0,0 +1,364 @@
+//
+// immer: immutable data structures for C++
+// Copyright (C) 2016, 2017, 2018 Juan Pedro Bolivar Puente
+//
+// This software is distributed under the Boost Software License, Version 1.0.
+// See accompanying file LICENSE or copy at http://boost.org/LICENSE_1_0.txt
+//
+
+#include "extra/fuzzer/fuzzer_input.hpp"
+#include <array>
+#include <catch.hpp>
+#include <immer/flex_vector.hpp>
+#include <immer/flex_vector_transient.hpp>
+#include <immer/heap/gc_heap.hpp>
+#include <immer/refcount/no_refcount_policy.hpp>
+#include <iostream>
+
+#define IMMER_FUZZED_TRACE_ENABLE 0
+
+#if IMMER_FUZZED_TRACE_ENABLE
+#define IMMER_FUZZED_TRACE(...) std::cout << __VA_ARGS__ << std::endl;
+#else
+#define IMMER_FUZZED_TRACE(...)
+#endif
+
+using gc_memory = immer::memory_policy<immer::heap_policy<immer::gc_heap>,
+                                       immer::no_refcount_policy,
+                                       immer::gc_transience_policy,
+                                       false>;
+
+namespace {
+
+template <std::size_t VarCount = 4, unsigned Bits = 2>
+int run_input(const std::uint8_t* data, std::size_t size)
+{
+    using vector_t    = immer::flex_vector<int, gc_memory, Bits, Bits>;
+    using transient_t = typename vector_t::transient_type;
+    using size_t      = std::uint8_t;
+
+    auto vs = std::array<vector_t, VarCount>{};
+    auto ts = std::array<transient_t, VarCount>{};
+
+#if IMMER_FUZZED_TRACE_ENABLE
+    std::cout << "/// new test run" << std::endl;
+    for (auto i = 0; i < VarCount; ++i)
+        std::cout << "auto v" << i << " = vector_t{};" << std::endl;
+    for (auto i = 0; i < VarCount; ++i)
+        std::cout << "auto t" << i << " = transient_t{};" << std::endl;
+#endif
+
+    auto is_valid_var = [&](auto idx) { return idx >= 0 && idx < VarCount; };
+    auto is_valid_var_neq = [](auto other) {
+        return [=](auto idx) {
+            return idx >= 0 && idx < VarCount && idx != other;
+        };
+    };
+    auto is_valid_index = [](auto& v) {
+        return [&](auto idx) { return idx >= 0 && idx < v.size(); };
+    };
+    auto is_valid_size = [](auto& v) {
+        return [&](auto idx) { return idx >= 0 && idx <= v.size(); };
+    };
+    auto can_concat = [](auto&& v1, auto&& v2) {
+        using size_type = decltype(v1.size());
+        auto max        = std::numeric_limits<size_type>::max() >> (Bits * 4);
+        return v1.size() < max && v2.size() < max;
+    };
+
+    return fuzzer_input{data, size}.run([&](auto& in) {
+        enum ops
+        {
+            op_transient,
+            op_persistent,
+            op_push_back,
+            op_update,
+            op_take,
+            op_drop,
+            op_concat,
+            op_push_back_mut,
+            op_update_mut,
+            op_take_mut,
+            op_drop_mut,
+            op_prepend_mut,
+            op_prepend_mut_move,
+            op_append_mut,
+            op_append_mut_move,
+        };
+        auto dst = read<std::uint8_t>(in, is_valid_var);
+        switch (read<char>(in)) {
+        case op_transient: {
+            auto src = read<std::uint8_t>(in, is_valid_var);
+            IMMER_FUZZED_TRACE("t" << +dst << " = v" << +src
+                                   << ".transient();");
+            ts[dst] = vs[src].transient();
+            break;
+        }
+        case op_persistent: {
+            auto src = read<std::uint8_t>(in, is_valid_var);
+            IMMER_FUZZED_TRACE("v" << +dst << " = t" << +src
+                                   << ".persistent();");
+            vs[dst] = ts[src].persistent();
+            break;
+        }
+        case op_push_back: {
+            auto src = read<std::uint8_t>(in, is_valid_var);
+            IMMER_FUZZED_TRACE("v" << +dst << " = v" << +src
+                                   << ".push_back(42);");
+            vs[dst] = vs[src].push_back(42);
+            break;
+        }
+        case op_update: {
+            auto src = read<std::uint8_t>(in, is_valid_var);
+            auto idx = read<size_t>(in, is_valid_index(vs[src]));
+            IMMER_FUZZED_TRACE("v" << +dst << " = v" << +src << ".update("
+                                   << +idx
+                                   << ", [] (auto x) { return x + 1; });");
+            vs[dst] = vs[src].update(idx, [](auto x) { return x + 1; });
+            break;
+        }
+        case op_take: {
+            auto src = read<std::uint8_t>(in, is_valid_var);
+            auto idx = read<size_t>(in, is_valid_size(vs[src]));
+            IMMER_FUZZED_TRACE("v" << +dst << " = v" << +src << ".take(" << +idx
+                                   << ");");
+            vs[dst] = vs[src].take(idx);
+            break;
+        }
+        case op_drop: {
+            auto src = read<std::uint8_t>(in, is_valid_var);
+            auto idx = read<size_t>(in, is_valid_size(vs[src]));
+            IMMER_FUZZED_TRACE("v" << +dst << " = v" << +src << ".take(" << +idx
+                                   << ");");
+            vs[dst] = vs[src].drop(idx);
+            break;
+        }
+        case op_concat: {
+            auto src  = read<std::uint8_t>(in, is_valid_var);
+            auto src2 = read<std::uint8_t>(in, is_valid_var);
+            if (can_concat(vs[src], vs[src2])) {
+                IMMER_FUZZED_TRACE("v" << +dst << " = v" << +src << " + v"
+                                       << +src2 << ";");
+                vs[dst] = vs[src] + vs[src2];
+            }
+            break;
+        }
+        case op_push_back_mut: {
+            IMMER_FUZZED_TRACE("t" << +dst << ".push_back(13);");
+            ts[dst].push_back(13);
+            break;
+        }
+        case op_update_mut: {
+            auto idx = read<size_t>(in, is_valid_index(ts[dst]));
+            IMMER_FUZZED_TRACE("t" << +dst << ".update(" << +idx
+                                   << ", [] (auto x) { return x + 1; });");
+            ts[dst].update(idx, [](auto x) { return x + 1; });
+            break;
+        }
+        case op_take_mut: {
+            auto idx = read<size_t>(in, is_valid_size(ts[dst]));
+            IMMER_FUZZED_TRACE("t" << +dst << ").take(" << +idx << ");");
+            ts[dst].take(idx);
+            break;
+        }
+        case op_prepend_mut: {
+            auto src = read<std::uint8_t>(in, is_valid_var_neq(dst));
+            if (can_concat(ts[dst], ts[src])) {
+                IMMER_FUZZED_TRACE("t" << +dst << ".prepend(t" << +src << ");");
+                ts[dst].prepend(ts[src]);
+            }
+            break;
+        }
+        case op_prepend_mut_move: {
+            auto src = read<std::uint8_t>(in, is_valid_var_neq(dst));
+            if (can_concat(ts[dst], ts[src])) {
+                IMMER_FUZZED_TRACE("t" << +dst << ".prepend(std::move(t" << +src
+                                       << "));"
+                                       << " t" << +src << " = {};");
+                ts[dst].prepend(std::move(ts[src]));
+                ts[src] = {};
+            }
+            break;
+        }
+        case op_append_mut: {
+            auto src = read<std::uint8_t>(in, is_valid_var_neq(dst));
+            if (can_concat(ts[dst], ts[src])) {
+                IMMER_FUZZED_TRACE("t" << +dst << ".append(t" << +src << ");");
+                ts[dst].append(ts[src]);
+            }
+            break;
+        }
+        case op_append_mut_move: {
+            auto src = read<std::uint8_t>(in, is_valid_var_neq(dst));
+            if (can_concat(ts[dst], ts[src])) {
+                IMMER_FUZZED_TRACE("t" << +dst << ".append(std::move(t" << +src
+                                       << "));"
+                                       << " t" << +src << " = {};");
+                ts[dst].append(std::move(ts[src]));
+                ts[src] = {};
+            }
+            break;
+        }
+        default:
+            break;
+        };
+        return true;
+    });
+}
+
+} // namespace
+
+TEST_CASE("bug: concatenating transients")
+{
+    // When concatenating two transients vectors the nodes from the
+    // argument become aliased in the result.  As such, we need to
+    // reset the identitiy of the argument.
+    SECTION("simplified")
+    {
+        using vector_t = immer::flex_vector<int, gc_memory, 2, 2>;
+        auto t0        = vector_t{}.transient();
+        t0.push_back(42);
+        t0.push_back(42);
+        t0.push_back(42);
+        t0.push_back(42);
+        t0.push_back(42);
+        t0.push_back(42);
+        auto t1 = t0;
+        t1.append(t0);
+        t1.append(t0);
+        t0.append(t1);
+        t1.append(t0);
+    }
+
+#if __GNUC__ != 9
+    SECTION("")
+    {
+        constexpr std::uint8_t input[] = {
+            0x2, 0x2, 0x2, 0x2, 0x29, 0x32, 0x0, 0x0,  0x2,  0x2,  0x2,
+            0x2, 0x2, 0x2, 0x2, 0x2,  0x2,  0x2, 0x2,  0x2,  0x6,  0x2,
+            0x2, 0x2, 0x2, 0x2, 0x6,  0x2,  0x2, 0x2,  0x0,  0x38, 0x2,
+            0x0, 0x0, 0x2, 0x2, 0xd,  0x0,  0x0, 0x3b, 0xff, 0x3a, 0x2,
+            0xd, 0xd, 0x2, 0x0, 0x0,  0x10, 0xe, 0x0,  0xd,  0x0,  0x0,
+            0x2, 0x2, 0xd, 0x0, 0x1,  0x5,
+        };
+        CHECK(run_input(input, sizeof(input)) == 0);
+    }
+#endif
+}
+
+TEST_CASE("bug: concatenating moved transients")
+{
+    // A moved from concatenated transient is totally smashed, we can
+    // not do anything with it but reasign...
+    SECTION("simplified")
+    {
+        using vector_t    = immer::flex_vector<int, gc_memory, 2, 2>;
+        using transient_t = typename vector_t::transient_type;
+        auto v0           = vector_t{};
+        auto t0           = transient_t{};
+        auto t2           = transient_t{};
+        v0                = v0.push_back(42);
+        t2                = v0.transient();
+        v0                = v0 + v0;
+        v0                = v0 + v0;
+        v0                = v0 + v0;
+        v0                = v0 + v0;
+        t0                = v0.transient();
+        t0                = v0.transient();
+        t0.append(std::move(t2));
+        t2 = {};
+        t2.append(std::move(t0));
+    }
+
+#if __GNUC__ != 9
+    SECTION("")
+    {
+        constexpr std::uint8_t input[] = {
+            0x0,  0x2, 0x0,  0x2,  0x0,  0x0,  0x0,  0x6,  0x0,  0x0,  0x0,
+            0x6,  0x0, 0x0,  0x0,  0x9d, 0x0,  0x6,  0x0,  0x0,  0x0,  0x6,
+            0x0,  0x0, 0x0,  0x9d, 0x28, 0x0,  0x0,  0x0,  0x0,  0x0,  0xf7,
+            0xc5, 0x0, 0xa,  0xa,  0x0,  0xfa, 0xe7, 0xff, 0xe7, 0xff, 0x0,
+            0xe,  0x2, 0x9,  0x0,  0x28, 0x2,  0xe,  0x0,  0x0,  0x2,  0xd,
+            0x0,  0x0, 0x28, 0x0,  0xd,  0x2,  0x5,  0x0,  0x2,
+        };
+        CHECK(run_input(input, sizeof(input)) == 0);
+    }
+#endif
+
+    SECTION("simplified")
+    {
+        using vector_t    = immer::flex_vector<int, gc_memory, 2, 2>;
+        using transient_t = typename vector_t::transient_type;
+        auto v0           = vector_t{};
+        auto t0           = transient_t{};
+        auto t2           = transient_t{};
+        v0                = v0.push_back(42);
+        v0                = v0.push_back(42);
+        v0                = v0 + v0;
+        t0                = v0.transient();
+        t2.prepend(std::move(t0));
+        t0 = {};
+        t0 = v0.transient();
+        t0.push_back(13);
+        t2.append(std::move(t0));
+        t0 = {};
+        t0 = v0.transient();
+        t0.push_back(13);
+        t2.prepend(std::move(t0));
+        t0 = {};
+    }
+
+#if __GNUC__ != 9
+    SECTION("")
+    {
+        return;
+        constexpr std::uint8_t input[] = {
+            0x0,  0x2, 0x0,  0x0,  0x2, 0xb7, 0x1, 0x36, 0x40, 0x0,  0x0,
+            0x0,  0x0, 0xb6, 0x0,  0x2, 0x0,  0x0, 0x6,  0xe,  0x0,  0x0,
+            0xfe, 0x0, 0x0,  0xff, 0x0, 0x2,  0xc, 0xff, 0xfc, 0x29, 0x0,
+            0x0,  0x0, 0x0,  0x0,  0x7, 0x2,  0xe, 0xff, 0xfc, 0x29, 0x0,
+            0x0,  0x0, 0x0,  0x0,  0x7, 0x3,  0x0, 0x0,  0x2,  0xc,  0x2,
+            0xc,  0x0, 0xd,  0x0,  0x0, 0x0,  0x0, 0x25, 0x6,
+        };
+        CHECK(run_input(input, sizeof(input)) == 0);
+    }
+#endif
+}
+
+TEST_CASE("bug: aegsdas")
+{
+    SECTION("simplified")
+    {
+        using vector_t    = immer::flex_vector<int, gc_memory, 2, 2>;
+        using transient_t = typename vector_t::transient_type;
+        auto v2           = vector_t{};
+        auto t0           = transient_t{};
+        auto t1           = transient_t{};
+        v2                = v2.push_back(42);
+        v2                = v2.push_back(42);
+        v2                = v2.push_back(42);
+        v2                = v2.push_back(42);
+        v2                = v2.push_back(42);
+        t0                = v2.transient();
+        t1.prepend(t0);
+        t1.prepend(t0);
+        t1.prepend(t0);
+        t0.prepend(std::move(t1));
+        t1 = {};
+    }
+
+#if __GNUC__ != 9
+    SECTION("")
+    {
+        constexpr std::uint8_t input[] = {
+            0xff, 0xff, 0x2, 0x2,  0x2, 0x2,  0x2,  0x2,  0x2,  0x2,  0x2,
+            0x82, 0x2,  0x2, 0x2,  0x2, 0x2,  0x2,  0x0,  0x0,  0x3d, 0x0,
+            0x0,  0x0,  0x2, 0x84, 0x0, 0x3b, 0x1,  0xb,  0x0,  0xa,  0x1,
+            0xb,  0x0,  0x0, 0x3,  0x2, 0x0,  0x3b, 0x1,  0xb,  0x1,  0x0,
+            0x0,  0xc,  0xb, 0x1,  0x8, 0xff, 0xff, 0xfc, 0xfd, 0x0,  0x3b,
+            0x3,  0x2,  0x0, 0x3b, 0x1, 0x9,  0x1,  0x3b,
+        };
+        CHECK(run_input(input, sizeof(input)) == 0);
+    }
+#endif
+}