Track mutant origins.

This is to enable input reduction (replacing corpus input with smaller mutants if the coverage matches) and corpus mutation stats for new scheduling methods.

PiperOrigin-RevId: 821630139

Author: xinhaoyuan

centipede/BUILD

@@ -369,7 +369,7 @@ cc_library(
     deps = [
         ":centipede_callbacks",
         ":environment",
-        ":mutation_input",
+        ":mutation_data",
         ":runner_result",
         ":stop",
         ":thread_pool",
@@ -442,6 +442,7 @@ cc_library(
         # used in centipede_runner.
         ":feature",
         ":execution_metadata",
+        ":mutation_data",
         ":shared_memory_blob_sequence",
         "@com_google_fuzztest//common:defs",
     ],
@@ -457,14 +458,15 @@ cc_library(
         # used in centipede_runner.
         ":shared_memory_blob_sequence",
         ":execution_metadata",
-        ":mutation_input",
+        ":mutation_data",
+        "@abseil-cpp//absl/types:span",
         "@com_google_fuzztest//common:defs",
     ],
 )
 
 cc_library(
-    name = "mutation_input",
-    hdrs = ["mutation_input.h"],
+    name = "mutation_data",
+    hdrs = ["mutation_data.h"],
     copts = DISABLE_SANCOV_COPTS,
     deps = [
         # This target must have a minimal set of dependencies since it is
@@ -483,8 +485,9 @@ cc_library(
     deps = [
         ":execution_metadata",
         ":knobs",
-        ":mutation_input",
+        ":mutation_data",
         "@abseil-cpp//absl/base:nullability",
+        "@abseil-cpp//absl/types:span",
         "@com_google_fuzztest//common:defs",
     ],
 )
@@ -627,7 +630,7 @@ cc_library(
         ":control_flow",
         ":environment",
         ":fuzztest_mutator",
-        ":mutation_input",
+        ":mutation_data",
         ":runner_request",
         ":runner_result",
         ":shared_memory_blob_sequence",
@@ -641,6 +644,7 @@ cc_library(
         "@abseil-cpp//absl/strings",
         "@abseil-cpp//absl/synchronization",
         "@abseil-cpp//absl/time",
+        "@abseil-cpp//absl/types:span",
         "@com_google_fuzztest//common:blob_file",
         "@com_google_fuzztest//common:defs",
         "@com_google_fuzztest//common:hash",
@@ -719,7 +723,7 @@ cc_library(
         ":environment",
         ":feature",
         ":feature_set",
-        ":mutation_input",
+        ":mutation_data",
         ":pc_info",
         ":runner_result",
         ":rusage_profiler",
@@ -737,6 +741,7 @@ cc_library(
         "@abseil-cpp//absl/strings:str_format",
         "@abseil-cpp//absl/synchronization",
         "@abseil-cpp//absl/time",
+        "@abseil-cpp//absl/types:span",
         "@com_google_fuzztest//common:blob_file",
         "@com_google_fuzztest//common:defs",
         "@com_google_fuzztest//common:hash",
@@ -855,11 +860,12 @@ cc_library(
     deps = [
         ":centipede_callbacks",
         ":environment",
-        ":mutation_input",
+        ":mutation_data",
         ":runner_result",
         ":stop",
         "@abseil-cpp//absl/status",
         "@abseil-cpp//absl/status:statusor",
+        "@abseil-cpp//absl/types:span",
         "@com_google_fuzztest//common:defs",
         "@com_google_fuzztest//common:logging",
     ],
@@ -873,7 +879,7 @@ cc_library(
         ":byte_array_mutator",
         ":execution_metadata",
         ":knobs",
-        ":mutation_input",
+        ":mutation_data",
         "@abseil-cpp//absl/random",
         "@abseil-cpp//absl/types:span",
         "@com_google_fuzztest//common:defs",
@@ -941,6 +947,7 @@ cc_library(
     name = "runner_cmp_trace",
     hdrs = ["runner_cmp_trace.h"],
     copts = DISABLE_SANCOV_COPTS,
+    deps = ["@abseil-cpp//absl/base:core_headers"],
 )
 
 # Library for manipulating centipede runner flags. This is not used by the
@@ -958,6 +965,7 @@ cc_library(
     hdrs = ["dispatcher.h"],
     deps = [
         ":execution_metadata",
+        ":mutation_data",
         ":runner_request",
         ":runner_result",
         ":shared_memory_blob_sequence",
@@ -1033,7 +1041,7 @@ RUNNER_DEPS = [
     ":foreach_nonzero",
     ":int_utils",
     ":knobs",
-    ":mutation_input",
+    ":mutation_data",
     ":rolling_hash",
     ":runner_cmp_trace",
     ":runner_fork_server",
@@ -1098,8 +1106,9 @@ cc_library(
     linkstatic = True,  # Must be linked statically even when dynamic_mode=on.
     deps = [
         ":centipede_runner_no_main",
-        ":mutation_input",
+        ":mutation_data",
         "@abseil-cpp//absl/base:nullability",
+        "@abseil-cpp//absl/types:span",
         "@com_google_fuzztest//common:defs",
     ],
 )
@@ -1237,9 +1246,10 @@ cc_library(
         ":corpus",
         ":environment",
         ":feature",
-        ":mutation_input",
+        ":mutation_data",
         ":runner_result",
         ":util",
+        "@abseil-cpp//absl/types:span",
         "@com_google_fuzztest//common:defs",
         "@com_google_fuzztest//common:logging",
     ],
@@ -1279,6 +1289,7 @@ cc_test(
         ":centipede_callbacks",
         ":environment",
         ":runner_result",
+        "@abseil-cpp//absl/types:span",
         "@com_google_fuzztest//common:defs",
         "@googletest//:gtest_main",
     ],
@@ -1456,6 +1467,7 @@ cc_test(
         ":util",
         ":workdir",
         "@abseil-cpp//absl/base:nullability",
+        "@abseil-cpp//absl/types:span",
         "@com_google_fuzztest//common:defs",
         "@com_google_fuzztest//common:test_util",
         "@googletest//:gtest_main",
@@ -1509,6 +1521,7 @@ cc_test(
     deps = [
         ":execution_metadata",
         ":feature",
+        ":mutation_data",
         ":runner_result",
         ":shared_memory_blob_sequence",
         "@com_google_fuzztest//common:defs",
@@ -1518,10 +1531,10 @@ cc_test(
 )
 
 cc_test(
-    name = "mutation_input_test",
-    srcs = ["mutation_input_test.cc"],
+    name = "mutation_data_test",
+    srcs = ["mutation_data_test.cc"],
     deps = [
-        ":mutation_input",
+        ":mutation_data",
         "@com_google_fuzztest//common:defs",
         "@googletest//:gtest_main",
     ],
@@ -1534,7 +1547,7 @@ cc_test(
         ":byte_array_mutator",
         ":execution_metadata",
         ":knobs",
-        ":mutation_input",
+        ":mutation_data",
         ":runner_cmp_trace",
         "@abseil-cpp//absl/container:flat_hash_set",
         "@com_google_fuzztest//common:defs",
@@ -1567,7 +1580,7 @@ cc_test(
         ":execution_metadata",
         ":fuzztest_mutator",
         ":knobs",
-        ":mutation_input",
+        ":mutation_data",
         "@abseil-cpp//absl/container:flat_hash_set",
         "@abseil-cpp//absl/strings",
         "@com_google_fuzztest//common:defs",
@@ -1872,6 +1885,7 @@ cc_test(
         "@abseil-cpp//absl/container:flat_hash_map",
         "@abseil-cpp//absl/strings:str_format",
         "@abseil-cpp//absl/time",
+        "@abseil-cpp//absl/types:span",
         "@com_google_fuzztest//common:defs",
         "@com_google_fuzztest//common:hash",
         "@com_google_fuzztest//common:temp_dir",
@@ -1914,14 +1928,15 @@ cc_test(
         ":centipede_interface",
         ":environment",
         ":feature",
-        ":mutation_input",
+        ":mutation_data",
         ":runner_result",
         ":stop",
         ":util",
         ":workdir",
         "@abseil-cpp//absl/container:flat_hash_set",
         "@abseil-cpp//absl/strings",
         "@abseil-cpp//absl/time",
+        "@abseil-cpp//absl/types:span",
         "@com_google_fuzztest//common:defs",
         "@com_google_fuzztest//common:hash",
         "@com_google_fuzztest//common:logging",

centipede/byte_array_mutator.cc

@@ -22,9 +22,10 @@
 #include <utility>
 #include <vector>
 
+#include "absl/types/span.h"
 #include "./centipede/execution_metadata.h"
 #include "./centipede/knobs.h"
-#include "./centipede/mutation_input.h"
+#include "./centipede/mutation_data.h"
 #include "./common/defs.h"
 
 namespace fuzztest::internal {
@@ -321,27 +322,29 @@ void ByteArrayMutator::CrossOver(ByteArray &data, const ByteArray &other) {
 // TODO(kcc): add tests with different values of knobs.
 const KnobId knob_mutate_or_crossover = Knobs::NewId("mutate_or_crossover");
 
-std::vector<ByteArray> ByteArrayMutator::MutateMany(
-    const std::vector<MutationInputRef> &inputs, size_t num_mutants) {
+std::vector<Mutant> ByteArrayMutator::MutateMany(
+    absl::Span<const MutationInputRef> inputs, size_t num_mutants) {
   if (inputs.empty()) abort();
   // TODO(xinhaoyuan): Consider metadata in other inputs instead of always the
   // first one.
   SetMetadata(inputs[0].metadata != nullptr ? *inputs[0].metadata
                                             : ExecutionMetadata());
   size_t num_inputs = inputs.size();
-  std::vector<ByteArray> mutants;
+  std::vector<Mutant> mutants;
   mutants.reserve(num_mutants);
   for (size_t i = 0; i < num_mutants; ++i) {
-    auto mutant = inputs[rng_() % num_inputs].data;
-    if (mutant.size() <= max_len_ &&
+    Mutant mutant;
+    mutant.origin = rng_() % num_inputs;
+    mutant.data = inputs[mutant.origin].data;
+    if (mutant.data.size() <= max_len_ &&
         knobs_.GenerateBool(knob_mutate_or_crossover, rng_())) {
       // Do crossover only if the mutant is not over the max_len_.
       // Perform crossover with some other input. It may be the same input.
       const auto &other_input = inputs[rng_() % num_inputs].data;
-      CrossOver(mutant, other_input);
+      CrossOver(mutant.data, other_input);
     } else {
       // Perform mutation.
-      Mutate(mutant);
+      Mutate(mutant.data);
     }
     mutants.push_back(std::move(mutant));
   }

centipede/byte_array_mutator.h

@@ -23,17 +23,18 @@
 #include <vector>
 
 #include "absl/base/nullability.h"
+#include "absl/types/span.h"
 #include "./centipede/execution_metadata.h"
 #include "./centipede/knobs.h"
-#include "./centipede/mutation_input.h"
+#include "./centipede/mutation_data.h"
 #include "./common/defs.h"
 
 namespace fuzztest::internal {
 
 // A simple class representing an array of up to kMaxEntrySize bytes.
 class DictEntry {
  public:
-  static constexpr uint8_t kMaxEntrySize = 16;
+  static constexpr uint8_t kMaxEntrySize = 128;
 
   explicit DictEntry(ByteSpan bytes)
       : bytes_{},  // initialize bytes_ to all zeros
@@ -108,8 +109,8 @@ class ByteArrayMutator {
   }
 
   // Takes non-empty `inputs` and produces `num_mutants` mutants.
-  std::vector<ByteArray> MutateMany(const std::vector<MutationInputRef> &inputs,
-                                    size_t num_mutants);
+  std::vector<Mutant> MutateMany(absl::Span<const MutationInputRef> inputs,
+                                 size_t num_mutants);
 
   using CrossOverFn = void (ByteArrayMutator::*)(ByteArray &,
                                                  const ByteArray &);