The C based gRPC (C++, Python, Ruby, Objective-C, PHP, C#) https://grpc.io/
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

715 lines
26 KiB

// Copyright 2015 gRPC authors.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
10 years ago
#ifndef GRPC_TEST_CORE_END2END_END2END_TESTS_H
#define GRPC_TEST_CORE_END2END_END2END_TESTS_H
10 years ago
#include <stdint.h>
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
#include <stdio.h>
#include <algorithm>
#include <functional>
#include <initializer_list>
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
#include <map>
#include <memory>
#include <string>
#include <utility>
#include <vector>
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
#include "absl/functional/any_invocable.h"
#include "absl/log/check.h"
#include "absl/memory/memory.h"
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
#include "absl/meta/type_traits.h"
[end2end] Explain failures a little better (#33621) I'd been adding the following stanza regularly to debug flakes/fuzz failures: ``` Expect(1, CoreEnd2endTest::MaybePerformAction{[&](bool success) { Crash(absl::StrCat( "Unexpected completion of client side call: success=", success ? "true" : "false", " status=", server_status.ToString(), " initial_md=", server_initial_metadata.ToString())); }}); ``` it was helpful because it indicated why a call batch finished successfully and helped quickly identify next steps. It occurred to me however that this would better be done inside of the framework, and for *all* ops that have outputs, so this PR does just that. Any time a batch with an op that outputs information finishes successfully but unexpectedly we now display those outputs in human readable form in the error message. Sample output: ``` [ RUN ] CorpusExamples/FuzzerCorpusTest.RunOneExample/0 RUN TEST: Http2SingleHopTest.SimpleDelayedRequestShort/Chttp2SimpleSslFullstack E0101 00:00:05.000000000 396633 simple_delayed_request.cc:37] Create client side call E0101 00:00:05.000000000 396633 simple_delayed_request.cc:41] Start initial batch E0101 00:00:05.000000000 396633 simple_delayed_request.cc:47] Start server E0101 00:00:05.000000000 396633 cq_verifier.cc:364] Verify tag(101)-✅ for 600000ms test/core/end2end/cq_verifier.cc:316: Unexpected event: OP_COMPLETE: tag:0x1 OK with: incoming_metadata: {} status_on_client: status=4 msg=Deadline Exceeded trailing_metadata={} checked @ test/core/end2end/tests/simple_delayed_request.cc:51 expected: test/core/end2end/tests/simple_delayed_request.cc:50: tag(101) success=true ```
1 year ago
#include "absl/strings/str_cat.h"
#include "absl/strings/string_view.h"
#include "absl/types/optional.h"
#include "absl/types/variant.h"
#include "gtest/gtest.h"
#include <grpc/byte_buffer.h>
#include <grpc/compression.h>
#include <grpc/credentials.h>
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
#include <grpc/event_engine/event_engine.h>
10 years ago
#include <grpc/grpc.h>
#include <grpc/grpc_security.h>
#include <grpc/impl/propagation_bits.h>
#include <grpc/status.h>
#include <grpc/support/alloc.h>
#include <grpc/support/log.h>
#include <grpc/support/time.h>
#include "src/core/lib/channel/channel_args.h"
#include "src/core/lib/gprpp/bitset.h"
#include "src/core/lib/gprpp/debug_location.h"
#include "src/core/lib/gprpp/time.h"
#include "src/core/lib/slice/slice.h"
#include "src/core/lib/slice/slice_internal.h"
#include "src/core/lib/surface/call_test_only.h"
#include "src/core/lib/surface/channel.h"
#include "test/core/call/batch_builder.h"
#include "test/core/end2end/cq_verifier.h"
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
#include "test/core/event_engine/event_engine_test_utils.h"
#include "test/core/test_util/test_config.h"
10 years ago
// Test feature flags.
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
#define FEATURE_MASK_DOES_NOT_SUPPORT_RETRY (1 << 0)
#define FEATURE_MASK_SUPPORTS_HOSTNAME_VERIFICATION (1 << 1)
// Feature mask supports call credentials with a minimum security level of
// GRPC_PRIVACY_AND_INTEGRITY.
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
#define FEATURE_MASK_SUPPORTS_PER_CALL_CREDENTIALS (1 << 2)
// Feature mask supports call credentials with a minimum security level of
// GRPC_SECURTITY_NONE.
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
#define FEATURE_MASK_SUPPORTS_PER_CALL_CREDENTIALS_LEVEL_INSECURE (1 << 3)
#define FEATURE_MASK_SUPPORTS_REQUEST_PROXYING (1 << 4)
#define FEATURE_MASK_SUPPORTS_CLIENT_CHANNEL (1 << 5)
#define FEATURE_MASK_IS_HTTP2 (1 << 6)
#define FEATURE_MASK_ENABLES_TRACES (1 << 7)
#define FEATURE_MASK_1BYTE_AT_A_TIME (1 << 8)
#define FEATURE_MASK_DOES_NOT_SUPPORT_WRITE_BUFFERING (1 << 9)
#define FEATURE_MASK_DOES_NOT_SUPPORT_CLIENT_HANDSHAKE_COMPLETE_FIRST (1 << 10)
#define FEATURE_MASK_IS_MINSTACK (1 << 11)
#define FEATURE_MASK_IS_SECURE (1 << 12)
#define FEATURE_MASK_DO_NOT_FUZZ (1 << 13)
// Exclude this fixture from experiment runs
#define FEATURE_MASK_EXCLUDE_FROM_EXPERIMENT_RUNS (1 << 14)
#define FEATURE_MASK_IS_CALL_V3 (1 << 15)
10 years ago
#define FAIL_AUTH_CHECK_SERVER_ARG_NAME "fail_auth_check"
namespace grpc_core {
extern bool g_is_fuzzing_core_e2e_tests;
class CoreTestFixture {
public:
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
virtual ~CoreTestFixture() = default;
virtual grpc_server* MakeServer(
const ChannelArgs& args, grpc_completion_queue* cq,
absl::AnyInvocable<void(grpc_server*)>& pre_server_start) = 0;
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
virtual grpc_channel* MakeClient(const ChannelArgs& args,
grpc_completion_queue* cq) = 0;
10 years ago
};
Slice RandomSlice(size_t length);
Slice RandomBinarySlice(size_t length);
struct CoreTestConfiguration {
// A descriptive name for this test fixture.
const char* name;
// Which features are supported by this fixture. See feature flags above.
uint32_t feature_mask;
// If the call host is setup by the fixture (for example, via the
// GRPC_SSL_TARGET_NAME_OVERRIDE_ARG channel arg), which value should the
// test expect to find in call_details.host
const char* overridden_call_host;
std::function<std::unique_ptr<CoreTestFixture>(
const ChannelArgs& client_args, const ChannelArgs& server_args)>
create_fixture;
10 years ago
};
// Base class for e2e tests.
//
// Initialization:
//
// At the start of a test, nothing is initialized. CoreConfiguration is reset,
// and there's certainly no server nor client.
// A test can then register whatever builders it wants into the
// CoreConfiguration and have them picked up. If it does not, it will get the
// default CoreConfiguration.
//
// The test may choose to then create a client and server with InitClient() and
// InitServer(). It does not matter which order they are called, nor whether one
// or both are called. It's necessary to call these if the test demands that
// non-empty channel args should be passed to either the client or server.
//
// If a test does not call InitClient() or InitServer(), then upon the first
// call to either NewClientCall() or NewServerCall(), the client and server will
// be instantiated - this saves substantial boilerplate in the most common case
// for our tests.
//
// Notes:
// - older compilers fail matching absl::string_view with some gmock matchers on
// older compilers, and it's tremendously convenient to be able to do so. So
// we use std::string for return types here - performance isn't particularly
// important, so an extra copy is fine.
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
class CoreEnd2endTest : public ::testing::Test {
public:
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
void TestInfrastructureSetParam(const CoreTestConfiguration* param) {
param_ = param;
}
const CoreTestConfiguration* GetParam() { return param_; }
void SetUp() override;
void TearDown() override;
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
virtual void RunTest() = 0;
void SetCqVerifierStepFn(
absl::AnyInvocable<
void(grpc_event_engine::experimental::EventEngine::Duration) const>
step_fn) {
step_fn_ = std::move(step_fn);
}
void SetQuiesceEventEngine(
absl::AnyInvocable<
void(std::shared_ptr<grpc_event_engine::experimental::EventEngine>&&)>
quiesce_event_engine) {
quiesce_event_engine_ = std::move(quiesce_event_engine);
}
class Call;
struct RegisteredCall {
void* p;
};
// Safe notification to use for core e2e tests.
// Since when we're fuzzing we don't run background threads, the normal
// Notification type isn't safe to wait on (for some background timer to fire
// for instance...), consequently we need to use this.
class TestNotification {
public:
explicit TestNotification(CoreEnd2endTest* test) : test_(test) {}
void WaitForNotificationWithTimeout(absl::Duration wait_time) {
if (g_is_fuzzing_core_e2e_tests) {
Timestamp end = Timestamp::Now() + Duration::NanosecondsRoundUp(
ToInt64Nanoseconds(wait_time));
while (true) {
if (base_.HasBeenNotified()) return;
auto now = Timestamp::Now();
if (now >= end) return;
test_->step_fn_(now - end);
}
} else {
base_.WaitForNotificationWithTimeout(wait_time);
}
}
void Notify() { base_.Notify(); }
private:
Notification base_;
CoreEnd2endTest* const test_;
};
// CallBuilder - results in a call to either grpc_channel_create_call or
// grpc_channel_create_registered_call.
// Affords a fluent interface to specify optional arguments.
class ClientCallBuilder {
public:
ClientCallBuilder(CoreEnd2endTest& test, std::string method)
: test_(test),
call_selector_(UnregisteredCall{std::move(method), absl::nullopt}) {}
ClientCallBuilder(CoreEnd2endTest& test, RegisteredCall registered_call)
: test_(test), call_selector_(registered_call.p) {}
// Specify the host (otherwise nullptr is passed)
ClientCallBuilder& Host(std::string host) {
absl::get<UnregisteredCall>(call_selector_).host = std::move(host);
return *this;
}
// Specify the timeout (otherwise gpr_inf_future is passed) - this time is
// scaled according to the test environment.
ClientCallBuilder& Timeout(Duration timeout) {
if (timeout == Duration::Infinity()) {
deadline_ = gpr_inf_future(GPR_CLOCK_REALTIME);
return *this;
}
deadline_ = grpc_timeout_milliseconds_to_deadline(timeout.millis());
return *this;
}
// Finally create the call.
Call Create();
private:
CoreEnd2endTest& test_;
struct UnregisteredCall {
std::string method;
absl::optional<std::string> host;
};
absl::variant<void*, UnregisteredCall> call_selector_;
grpc_call* parent_call_ = nullptr;
uint32_t propagation_mask_ = GRPC_PROPAGATE_DEFAULTS;
gpr_timespec deadline_ = gpr_inf_future(GPR_CLOCK_REALTIME);
};
// Wrapper around a grpc_call.
// Instantiated by ClientCallBuilder via NewClientCall for client calls.
// Wrapped by IncomingCall for server calls.
class Call {
public:
[end2end] Explain failures a little better (#33621) I'd been adding the following stanza regularly to debug flakes/fuzz failures: ``` Expect(1, CoreEnd2endTest::MaybePerformAction{[&](bool success) { Crash(absl::StrCat( "Unexpected completion of client side call: success=", success ? "true" : "false", " status=", server_status.ToString(), " initial_md=", server_initial_metadata.ToString())); }}); ``` it was helpful because it indicated why a call batch finished successfully and helped quickly identify next steps. It occurred to me however that this would better be done inside of the framework, and for *all* ops that have outputs, so this PR does just that. Any time a batch with an op that outputs information finishes successfully but unexpectedly we now display those outputs in human readable form in the error message. Sample output: ``` [ RUN ] CorpusExamples/FuzzerCorpusTest.RunOneExample/0 RUN TEST: Http2SingleHopTest.SimpleDelayedRequestShort/Chttp2SimpleSslFullstack E0101 00:00:05.000000000 396633 simple_delayed_request.cc:37] Create client side call E0101 00:00:05.000000000 396633 simple_delayed_request.cc:41] Start initial batch E0101 00:00:05.000000000 396633 simple_delayed_request.cc:47] Start server E0101 00:00:05.000000000 396633 cq_verifier.cc:364] Verify tag(101)-✅ for 600000ms test/core/end2end/cq_verifier.cc:316: Unexpected event: OP_COMPLETE: tag:0x1 OK with: incoming_metadata: {} status_on_client: status=4 msg=Deadline Exceeded trailing_metadata={} checked @ test/core/end2end/tests/simple_delayed_request.cc:51 expected: test/core/end2end/tests/simple_delayed_request.cc:50: tag(101) success=true ```
1 year ago
Call(grpc_call* call, CoreEnd2endTest* test) : call_(call), test_(test) {}
Call(const Call&) = delete;
Call& operator=(const Call&) = delete;
[end2end] Explain failures a little better (#33621) I'd been adding the following stanza regularly to debug flakes/fuzz failures: ``` Expect(1, CoreEnd2endTest::MaybePerformAction{[&](bool success) { Crash(absl::StrCat( "Unexpected completion of client side call: success=", success ? "true" : "false", " status=", server_status.ToString(), " initial_md=", server_initial_metadata.ToString())); }}); ``` it was helpful because it indicated why a call batch finished successfully and helped quickly identify next steps. It occurred to me however that this would better be done inside of the framework, and for *all* ops that have outputs, so this PR does just that. Any time a batch with an op that outputs information finishes successfully but unexpectedly we now display those outputs in human readable form in the error message. Sample output: ``` [ RUN ] CorpusExamples/FuzzerCorpusTest.RunOneExample/0 RUN TEST: Http2SingleHopTest.SimpleDelayedRequestShort/Chttp2SimpleSslFullstack E0101 00:00:05.000000000 396633 simple_delayed_request.cc:37] Create client side call E0101 00:00:05.000000000 396633 simple_delayed_request.cc:41] Start initial batch E0101 00:00:05.000000000 396633 simple_delayed_request.cc:47] Start server E0101 00:00:05.000000000 396633 cq_verifier.cc:364] Verify tag(101)-✅ for 600000ms test/core/end2end/cq_verifier.cc:316: Unexpected event: OP_COMPLETE: tag:0x1 OK with: incoming_metadata: {} status_on_client: status=4 msg=Deadline Exceeded trailing_metadata={} checked @ test/core/end2end/tests/simple_delayed_request.cc:51 expected: test/core/end2end/tests/simple_delayed_request.cc:50: tag(101) success=true ```
1 year ago
Call(Call&& other) noexcept
: call_(std::exchange(other.call_, nullptr)),
test_(std::exchange(other.test_, nullptr)) {}
~Call() {
if (call_ != nullptr) grpc_call_unref(call_);
}
// Construct a batch with a tag - upon destruction of the BatchBuilder the
// operation will occur.
BatchBuilder NewBatch(int tag) {
return BatchBuilder(call_, &test_->cq_verifier(), tag);
}
// Cancel the call
void Cancel() { grpc_call_cancel(call_, nullptr); }
void CancelWithStatus(grpc_status_code status, const char* message) {
grpc_call_cancel_with_status(call_, status, message, nullptr);
}
// Access the peer structure (returns a string that can be matched, etc) -
// or nullopt if grpc_call_get_peer returns nullptr.
absl::optional<std::string> GetPeer() {
char* peer = grpc_call_get_peer(call_);
if (peer == nullptr) return absl::nullopt;
std::string result(peer);
gpr_free(peer);
return result;
}
// Set call credentials.
// Takes ownership of creds.
void SetCredentials(grpc_call_credentials* creds) {
EXPECT_EQ(grpc_call_set_credentials(call_, creds), GRPC_CALL_OK);
grpc_call_credentials_release(creds);
}
// Retrieve the auth context.
std::unique_ptr<grpc_auth_context, void (*)(grpc_auth_context*)>
GetAuthContext() {
return std::unique_ptr<grpc_auth_context, void (*)(grpc_auth_context*)>(
grpc_call_auth_context(call_), grpc_auth_context_release);
}
grpc_call** call_ptr() { return &call_; }
grpc_call* c_call() const { return call_; }
private:
grpc_call* call_ = nullptr;
[end2end] Explain failures a little better (#33621) I'd been adding the following stanza regularly to debug flakes/fuzz failures: ``` Expect(1, CoreEnd2endTest::MaybePerformAction{[&](bool success) { Crash(absl::StrCat( "Unexpected completion of client side call: success=", success ? "true" : "false", " status=", server_status.ToString(), " initial_md=", server_initial_metadata.ToString())); }}); ``` it was helpful because it indicated why a call batch finished successfully and helped quickly identify next steps. It occurred to me however that this would better be done inside of the framework, and for *all* ops that have outputs, so this PR does just that. Any time a batch with an op that outputs information finishes successfully but unexpectedly we now display those outputs in human readable form in the error message. Sample output: ``` [ RUN ] CorpusExamples/FuzzerCorpusTest.RunOneExample/0 RUN TEST: Http2SingleHopTest.SimpleDelayedRequestShort/Chttp2SimpleSslFullstack E0101 00:00:05.000000000 396633 simple_delayed_request.cc:37] Create client side call E0101 00:00:05.000000000 396633 simple_delayed_request.cc:41] Start initial batch E0101 00:00:05.000000000 396633 simple_delayed_request.cc:47] Start server E0101 00:00:05.000000000 396633 cq_verifier.cc:364] Verify tag(101)-✅ for 600000ms test/core/end2end/cq_verifier.cc:316: Unexpected event: OP_COMPLETE: tag:0x1 OK with: incoming_metadata: {} status_on_client: status=4 msg=Deadline Exceeded trailing_metadata={} checked @ test/core/end2end/tests/simple_delayed_request.cc:51 expected: test/core/end2end/tests/simple_delayed_request.cc:50: tag(101) success=true ```
1 year ago
CoreEnd2endTest* test_;
};
// Wrapper around a server call.
class IncomingCall {
public:
IncomingCall(CoreEnd2endTest& test, int tag);
IncomingCall(CoreEnd2endTest& test, void* method, IncomingMessage* message,
int tag);
IncomingCall(const IncomingCall&) = delete;
IncomingCall& operator=(const IncomingCall&) = delete;
IncomingCall(IncomingCall&&) noexcept = default;
// Construct a batch with a tag - upon destruction of the BatchBuilder the
// operation will occur. Must have received the call first!
BatchBuilder NewBatch(int tag) { return impl_->call.NewBatch(tag); }
void Cancel() { impl_->call.Cancel(); }
// Return the method being called.
std::string method() const {
return std::string(StringViewFromSlice(impl_->call_details.method));
}
// Return the host being called.
std::string host() const {
return std::string(StringViewFromSlice(impl_->call_details.host));
}
// Return some initial metadata.
absl::optional<std::string> GetInitialMetadata(absl::string_view key) const;
// Return the peer address.
absl::optional<std::string> GetPeer() { return impl_->call.GetPeer(); }
// Return the auth context.
std::unique_ptr<grpc_auth_context, void (*)(grpc_auth_context*)>
GetAuthContext() {
return impl_->call.GetAuthContext();
}
// Return the underlying C call object
grpc_call* c_call() { return impl_->call.c_call(); }
// Return the encodings accepted by the peer.
BitSet<GRPC_COMPRESS_ALGORITHMS_COUNT> GetEncodingsAcceptedByPeer() {
return BitSet<GRPC_COMPRESS_ALGORITHMS_COUNT>::FromInt(
grpc_call_test_only_get_encodings_accepted_by_peer(c_call()));
}
private:
struct Impl {
[end2end] Explain failures a little better (#33621) I'd been adding the following stanza regularly to debug flakes/fuzz failures: ``` Expect(1, CoreEnd2endTest::MaybePerformAction{[&](bool success) { Crash(absl::StrCat( "Unexpected completion of client side call: success=", success ? "true" : "false", " status=", server_status.ToString(), " initial_md=", server_initial_metadata.ToString())); }}); ``` it was helpful because it indicated why a call batch finished successfully and helped quickly identify next steps. It occurred to me however that this would better be done inside of the framework, and for *all* ops that have outputs, so this PR does just that. Any time a batch with an op that outputs information finishes successfully but unexpectedly we now display those outputs in human readable form in the error message. Sample output: ``` [ RUN ] CorpusExamples/FuzzerCorpusTest.RunOneExample/0 RUN TEST: Http2SingleHopTest.SimpleDelayedRequestShort/Chttp2SimpleSslFullstack E0101 00:00:05.000000000 396633 simple_delayed_request.cc:37] Create client side call E0101 00:00:05.000000000 396633 simple_delayed_request.cc:41] Start initial batch E0101 00:00:05.000000000 396633 simple_delayed_request.cc:47] Start server E0101 00:00:05.000000000 396633 cq_verifier.cc:364] Verify tag(101)-✅ for 600000ms test/core/end2end/cq_verifier.cc:316: Unexpected event: OP_COMPLETE: tag:0x1 OK with: incoming_metadata: {} status_on_client: status=4 msg=Deadline Exceeded trailing_metadata={} checked @ test/core/end2end/tests/simple_delayed_request.cc:51 expected: test/core/end2end/tests/simple_delayed_request.cc:50: tag(101) success=true ```
1 year ago
explicit Impl(CoreEnd2endTest* test) : call(nullptr, test) {
grpc_call_details_init(&call_details);
grpc_metadata_array_init(&request_metadata);
}
~Impl() {
grpc_call_details_destroy(&call_details);
grpc_metadata_array_destroy(&request_metadata);
}
[end2end] Explain failures a little better (#33621) I'd been adding the following stanza regularly to debug flakes/fuzz failures: ``` Expect(1, CoreEnd2endTest::MaybePerformAction{[&](bool success) { Crash(absl::StrCat( "Unexpected completion of client side call: success=", success ? "true" : "false", " status=", server_status.ToString(), " initial_md=", server_initial_metadata.ToString())); }}); ``` it was helpful because it indicated why a call batch finished successfully and helped quickly identify next steps. It occurred to me however that this would better be done inside of the framework, and for *all* ops that have outputs, so this PR does just that. Any time a batch with an op that outputs information finishes successfully but unexpectedly we now display those outputs in human readable form in the error message. Sample output: ``` [ RUN ] CorpusExamples/FuzzerCorpusTest.RunOneExample/0 RUN TEST: Http2SingleHopTest.SimpleDelayedRequestShort/Chttp2SimpleSslFullstack E0101 00:00:05.000000000 396633 simple_delayed_request.cc:37] Create client side call E0101 00:00:05.000000000 396633 simple_delayed_request.cc:41] Start initial batch E0101 00:00:05.000000000 396633 simple_delayed_request.cc:47] Start server E0101 00:00:05.000000000 396633 cq_verifier.cc:364] Verify tag(101)-✅ for 600000ms test/core/end2end/cq_verifier.cc:316: Unexpected event: OP_COMPLETE: tag:0x1 OK with: incoming_metadata: {} status_on_client: status=4 msg=Deadline Exceeded trailing_metadata={} checked @ test/core/end2end/tests/simple_delayed_request.cc:51 expected: test/core/end2end/tests/simple_delayed_request.cc:50: tag(101) success=true ```
1 year ago
Call call;
grpc_call_details call_details;
grpc_metadata_array request_metadata;
};
std::unique_ptr<Impl> impl_;
};
class ServerRegisteredMethod {
public:
ServerRegisteredMethod(
CoreEnd2endTest* test, absl::string_view name,
grpc_server_register_method_payload_handling payload_handling);
void* handle() { return *handle_; }
private:
std::shared_ptr<void*> handle_ = std::make_shared<void*>(nullptr);
};
ServerRegisteredMethod RegisterServerMethod(
absl::string_view name,
grpc_server_register_method_payload_handling payload_handling) {
return ServerRegisteredMethod(this, name, payload_handling);
}
// Begin construction of a client call.
ClientCallBuilder NewClientCall(std::string method) {
return ClientCallBuilder(*this, std::move(method));
}
ClientCallBuilder NewClientCall(RegisteredCall registered_method) {
return ClientCallBuilder(*this, registered_method);
}
// Request a call on the server - notifies `tag` when complete.
IncomingCall RequestCall(int tag) { return IncomingCall(*this, tag); }
// Request a call on the server - notifies `tag` when complete.
IncomingCall RequestRegisteredCall(ServerRegisteredMethod method, int tag) {
return IncomingCall(*this, method.handle(), nullptr, tag);
}
IncomingCall RequestRegisteredCall(ServerRegisteredMethod method,
IncomingMessage* message, int tag) {
return IncomingCall(*this, method.handle(), message, tag);
}
// Pull in CqVerifier types for ergonomics
using ExpectedResult = CqVerifier::ExpectedResult;
using Maybe = CqVerifier::Maybe;
using PerformAction = CqVerifier::PerformAction;
using MaybePerformAction = CqVerifier::MaybePerformAction;
using AnyStatus = CqVerifier::AnyStatus;
// Expect a tag with some result.
void Expect(int tag, ExpectedResult result, SourceLocation whence = {}) {
expectations_++;
[end2end] Explain failures a little better (#33621) I'd been adding the following stanza regularly to debug flakes/fuzz failures: ``` Expect(1, CoreEnd2endTest::MaybePerformAction{[&](bool success) { Crash(absl::StrCat( "Unexpected completion of client side call: success=", success ? "true" : "false", " status=", server_status.ToString(), " initial_md=", server_initial_metadata.ToString())); }}); ``` it was helpful because it indicated why a call batch finished successfully and helped quickly identify next steps. It occurred to me however that this would better be done inside of the framework, and for *all* ops that have outputs, so this PR does just that. Any time a batch with an op that outputs information finishes successfully but unexpectedly we now display those outputs in human readable form in the error message. Sample output: ``` [ RUN ] CorpusExamples/FuzzerCorpusTest.RunOneExample/0 RUN TEST: Http2SingleHopTest.SimpleDelayedRequestShort/Chttp2SimpleSslFullstack E0101 00:00:05.000000000 396633 simple_delayed_request.cc:37] Create client side call E0101 00:00:05.000000000 396633 simple_delayed_request.cc:41] Start initial batch E0101 00:00:05.000000000 396633 simple_delayed_request.cc:47] Start server E0101 00:00:05.000000000 396633 cq_verifier.cc:364] Verify tag(101)-✅ for 600000ms test/core/end2end/cq_verifier.cc:316: Unexpected event: OP_COMPLETE: tag:0x1 OK with: incoming_metadata: {} status_on_client: status=4 msg=Deadline Exceeded trailing_metadata={} checked @ test/core/end2end/tests/simple_delayed_request.cc:51 expected: test/core/end2end/tests/simple_delayed_request.cc:50: tag(101) success=true ```
1 year ago
cq_verifier().Expect(CqVerifier::tag(tag), std::move(result), whence);
}
// Step the system until expectations are met or until timeout is reached.
// If there are no expectations logged, then step for 1 second and verify that
// no events occur.
void Step(absl::optional<Duration> timeout = absl::nullopt,
SourceLocation whence = {}) {
if (expectations_ == 0) {
cq_verifier().VerifyEmpty(timeout.value_or(Duration::Seconds(1)), whence);
return;
}
expectations_ = 0;
cq_verifier().Verify(
timeout.value_or(g_is_fuzzing_core_e2e_tests ? Duration::Minutes(10)
: Duration::Seconds(10)),
whence);
}
// Initialize the client.
// If called, then InitServer must be called to create a server (otherwise one
// will be provided).
void InitClient(const ChannelArgs& args) {
initialized_ = true;
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
if (client_ != nullptr) ShutdownAndDestroyClient();
auto& f = fixture();
client_ = f.MakeClient(args, cq_);
CHECK_NE(client_, nullptr);
}
// Initialize the server.
// If called, then InitClient must be called to create a client (otherwise one
// will be provided).
void InitServer(const ChannelArgs& args) {
initialized_ = true;
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
if (server_ != nullptr) ShutdownAndDestroyServer();
auto& f = fixture();
server_ = f.MakeServer(args, cq_, pre_server_start_);
CHECK_NE(server_, nullptr);
}
// Remove the client.
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
void ShutdownAndDestroyClient() {
if (client_ == nullptr) return;
grpc_channel_destroy(client_);
client_ = nullptr;
}
// Shutdown the server; notify tag on completion.
void ShutdownServerAndNotify(int tag) {
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
grpc_server_shutdown_and_notify(server_, cq_, CqVerifier::tag(tag));
}
// Destroy the server.
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
void DestroyServer() {
if (server_ == nullptr) return;
grpc_server_destroy(server_);
server_ = nullptr;
}
// Shutdown then destroy the server.
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
void ShutdownAndDestroyServer() {
if (server_ == nullptr) return;
ShutdownServerAndNotify(-1);
Expect(-1, AnyStatus{});
Step();
DestroyServer();
}
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
// Cancel any calls on the server.
void CancelAllCallsOnServer() { grpc_server_cancel_all_calls(server_); }
// Ping the server from the client
void PingServerFromClient(int tag) {
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
grpc_channel_ping(client_, cq_, CqVerifier::tag(tag), nullptr);
}
// Register a call on the client, return its handle.
RegisteredCall RegisterCallOnClient(const char* method, const char* host) {
ForceInitialized();
return RegisteredCall{
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
grpc_channel_register_call(client_, method, host, nullptr)};
}
// Return the current connectivity state of the client.
grpc_connectivity_state CheckConnectivityState(bool try_to_connect) {
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
return grpc_channel_check_connectivity_state(client_, try_to_connect);
}
// Watch the connectivity state of the client.
void WatchConnectivityState(grpc_connectivity_state last_observed_state,
Duration deadline, int tag) {
grpc_channel_watch_connectivity_state(
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
client_, last_observed_state,
grpc_timeout_milliseconds_to_deadline(deadline.millis()), cq_,
CqVerifier::tag(tag));
}
// Return the client channel.
grpc_channel* client() {
ForceInitialized();
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
return client_;
}
// Return the server channel.
grpc_server* server() {
ForceInitialized();
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
return server_;
}
grpc_completion_queue* cq() {
ForceInitialized();
return cq_;
}
// Given a duration, return a timestamp that is that duration in the future -
// with dilation according to test environment (eg sanitizers)
Timestamp TimestampAfterDuration(Duration duration) {
return Timestamp::FromTimespecRoundUp(
grpc_timeout_milliseconds_to_deadline(duration.millis()));
}
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
void SetPostGrpcInitFunc(absl::AnyInvocable<void()> fn) {
CHECK(fixture_ == nullptr);
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
post_grpc_init_func_ = std::move(fn);
}
private:
void ForceInitialized();
CoreTestFixture& fixture() {
if (fixture_ == nullptr) {
grpc_init();
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
post_grpc_init_func_();
cq_ = grpc_completion_queue_create_for_next(nullptr);
fixture_ = GetParam()->create_fixture(ChannelArgs(), ChannelArgs());
}
return *fixture_;
}
CqVerifier& cq_verifier() {
if (cq_verifier_ == nullptr) {
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
fixture(); // ensure cq_ present
cq_verifier_ = absl::make_unique<CqVerifier>(
cq_,
g_is_fuzzing_core_e2e_tests ? CqVerifier::FailUsingGprCrashWithStdio
: CqVerifier::FailUsingGprCrash,
step_fn_ == nullptr
? nullptr
: absl::AnyInvocable<void(
grpc_event_engine::experimental::EventEngine::Duration)
const>(
[this](
grpc_event_engine::experimental::EventEngine::Duration
d) { step_fn_(d); }));
}
return *cq_verifier_;
}
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
const CoreTestConfiguration* param_ = nullptr;
std::unique_ptr<CoreTestFixture> fixture_;
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
grpc_completion_queue* cq_ = nullptr;
grpc_server* server_ = nullptr;
grpc_channel* client_ = nullptr;
std::unique_ptr<CqVerifier> cq_verifier_;
absl::AnyInvocable<void(grpc_server*)> pre_server_start_ = [](grpc_server*) {
};
int expectations_ = 0;
bool initialized_ = false;
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
absl::AnyInvocable<void()> post_grpc_init_func_ = []() {};
absl::AnyInvocable<void(
grpc_event_engine::experimental::EventEngine::Duration) const>
step_fn_ = nullptr;
absl::AnyInvocable<void(
std::shared_ptr<grpc_event_engine::experimental::EventEngine>&&)>
quiesce_event_engine_ =
grpc_event_engine::experimental::WaitForSingleOwner;
};
// Define names for additional test suites.
// These make no changes to the actual class, but define new names to register
// tests against. Each new name gets a differing set of configurations in
// end2end_test_main.cc to customize the set of fixtures the tests run against.
// Test suite for tests that rely on a secure transport
class SecureEnd2endTest : public CoreEnd2endTest {};
// Test suite for tests that send rather large messages/metadata
class CoreLargeSendTest : public CoreEnd2endTest {};
// Test suite for tests that need a client channel
class CoreClientChannelTest : public CoreEnd2endTest {};
// Test suite for tests that require deadline handling
class CoreDeadlineTest : public CoreEnd2endTest {};
// Test suite for tests that require deadline handling
class CoreDeadlineSingleHopTest : public CoreEnd2endTest {};
// Test suite for http2 tests that only work over a single hop (unproxyable)
class Http2SingleHopTest : public CoreEnd2endTest {};
// Test suite for fullstack single hop http2 tests (require client channel)
class Http2FullstackSingleHopTest : public CoreEnd2endTest {};
// Test suite for tests that require retry features
class RetryTest : public CoreEnd2endTest {};
// Test suite for write buffering
class WriteBufferingTest : public CoreEnd2endTest {};
// Test suite for http2 tests
class Http2Test : public CoreEnd2endTest {};
// Test suite for http2 tests that require retry features
class RetryHttp2Test : public CoreEnd2endTest {};
// Test suite for tests that require resource quota
class ResourceQuotaTest : public CoreEnd2endTest {};
// Test suite for tests that require a transport that supports secure call
// credentials
class PerCallCredsTest : public CoreEnd2endTest {};
// Test suite for tests that require a transport that supports insecure call
// credentials
class PerCallCredsOnInsecureTest : public CoreEnd2endTest {};
// Test suite for tests that verify lack of logging in particular situations
class NoLoggingTest : public CoreEnd2endTest {};
// Test suite for tests that verify proxy authentication
class ProxyAuthTest : public CoreEnd2endTest {};
10 years ago
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
using MakeTestFn = absl::AnyInvocable<CoreEnd2endTest*(
const CoreTestConfiguration* config) const>;
class CoreEnd2endTestRegistry {
public:
CoreEnd2endTestRegistry(const CoreEnd2endTestRegistry&) = delete;
CoreEnd2endTestRegistry& operator=(const CoreEnd2endTestRegistry&) = delete;
static CoreEnd2endTestRegistry& Get() {
static CoreEnd2endTestRegistry* singleton = new CoreEnd2endTestRegistry;
return *singleton;
}
struct Test {
absl::string_view suite;
absl::string_view name;
const CoreTestConfiguration* config;
const MakeTestFn& make_test;
};
void RegisterTest(absl::string_view suite, absl::string_view name,
MakeTestFn make_test, SourceLocation where = {});
void RegisterSuite(absl::string_view suite,
std::vector<const CoreTestConfiguration*> configs,
SourceLocation where);
std::vector<Test> AllTests();
// Enforce passing a type so that we can check it exists (saves typos)
template <typename T>
absl::void_t<T> RegisterSuiteT(
absl::string_view suite,
std::vector<const CoreTestConfiguration*> configs,
SourceLocation where = {}) {
return RegisterSuite(suite, std::move(configs), where);
}
private:
CoreEnd2endTestRegistry() = default;
std::map<absl::string_view, std::vector<const CoreTestConfiguration*>>
suites_;
std::map<absl::string_view, std::map<absl::string_view, MakeTestFn>>
tests_by_suite_;
};
} // namespace grpc_core
// If this test fixture is being run under minstack, skip the test.
#define SKIP_IF_MINSTACK() \
if (GetParam()->feature_mask & FEATURE_MASK_IS_MINSTACK) \
GTEST_SKIP() << "Skipping test for minstack"
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
#define SKIP_IF_FUZZING() \
if (g_is_fuzzing_core_e2e_tests) GTEST_SKIP() << "Skipping test for fuzzing"
#define SKIP_IF_V3() \
if (GetParam()->feature_mask & FEATURE_MASK_IS_CALL_V3) { \
GTEST_SKIP() << "Disabled for initial v3 testing"; \
}
[fuzz] Turn core end2end tests into fuzzers (#33013) Add a new binary that runs all core end2end tests in fuzzing mode. In this mode FuzzingEventEngine is substituted for the default event engine. This means that time is simulated, as is IO. The FEE gets control of callback delays also. In our tests the `Step()` function becomes, instead of a single call to `completion_queue_next`, a series of calls to that function and `FuzzingEventEngine::Tick`, driving forward the event loop until progress can be made. PR guide: --- **New binaries** `core_end2end_test_fuzzer` - the new fuzzer itself `seed_end2end_corpus` - a tool that produces an interesting seed corpus **Config changes for safe fuzzing** The implementation tries to use the config fuzzing work we've previously deployed in api_fuzzer to fuzz across experiments. Since some experiments are far too experimental to be safe in such fuzzing (and this will always be the case): - a new flag is added to experiments to opt-out of this fuzzing - a new hook is added to the config system to allow variables to re-write their inputs before setting them during the fuzz **Event manager/IO changes** Changes are made to the event engine shims so that tcp_server_posix can run with a non-FD carrying EventEngine. These are in my mind a bit clunky, but they work and they're in code that we expect to delete in the medium term, so I think overall the approach is good. **Changes to time** A small tweak is made to fix a bug initializing time for fuzzers in time.cc - we were previously failing to initialize `g_process_epoch_cycles` **Changes to `Crash`** A version that prints to stdio is added so that we can reliably print a crash from the fuzzer. **Changes to CqVerifier** Hooks are added to allow the top level loop to hook the verification functions with a function that steps time between CQ polls. **Changes to end2end fixtures** State machinery moves from the fixture to the test infra, to keep the customizations for fuzzing or not in one place. This means that fixtures are now just client/server factories, which is overall nice. It did necessitate moving some bespoke machinery into h2_ssl_cert_test.cc - this file is beginning to be problematic in borrowing parts but not all of the e2e test machinery. Some future PR needs to solve this. A cq arg is added to the Make functions since the cq is now owned by the test and not the fixture. **Changes to test registration** `TEST_P` is replaced by `CORE_END2END_TEST` and our own test registry is used as a first depot for test information. The gtest version of these tests: queries that registry to manually register tests with gtest. This ultimately changes the name of our tests again (I think for the last time) - the new names are shorter and more readable, so I don't count this as a regression. The fuzzer version of these tests: constructs a database of fuzzable tests that it can consult to look up a particular suite/test/config combination specified by the fuzzer to fuzz against. This gives us a single fuzzer that can test all 3k-ish fuzzing ready tests and cross polinate configuration between them. **Changes to test config** The zero size registry stuff was causing some problems with the event engine feature macros, so instead I've removed those and used GTEST_SKIP in the problematic tests. I think that's the approach we move towards in the future. **Which tests are included** Configs that are compatible - those that do not do fd manipulation directly (these are incompatible with FuzzingEventEngine), and those that do not join threads on their shutdown path (as these are incompatible with our cq wait methodology). Each we can talk about in the future - fd manipulation would be a significant expansion of FuzzingEventEngine, and is probably not worth it, however many uses of background threads now should probably evolve to be EventEngine::Run calls in the future, and then would be trivially enabled in the fuzzers. Some tests currently fail in the fuzzing environment, a `SKIP_IF_FUZZING` macro is used for these few to disable them if in the fuzzing environment. We'll burn these down in the future. **Changes to fuzzing_event_engine** Changes are made to time: an exponential sweep forward is used now - this catches small time precision things early, but makes decade long timers (we have them) able to be used right now. In the future we'll just skip time forward to the next scheduled timer, but that approach doesn't yet work due to legacy timer system interactions. Changes to port assignment: we ensure that ports are legal numbers before assigning them via `grpc_pick_port_or_die`. A race condition between time checking and io is fixed. --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
#define CORE_END2END_TEST(suite, name) \
class CoreEnd2endTest_##suite##_##name : public grpc_core::suite { \
public: \
CoreEnd2endTest_##suite##_##name() {} \
void TestBody() override { RunTest(); } \
void RunTest() override; \
\
private: \
static grpc_core::CoreEnd2endTest* Run( \
const grpc_core::CoreTestConfiguration* config) { \
auto* test = new CoreEnd2endTest_##suite##_##name; \
test->TestInfrastructureSetParam(config); \
return test; \
} \
static int registered_; \
}; \
int CoreEnd2endTest_##suite##_##name::registered_ = \
(grpc_core::CoreEnd2endTestRegistry::Get().RegisterTest(#suite, #name, \
&Run), \
0); \
void CoreEnd2endTest_##suite##_##name::RunTest()
#define CORE_END2END_TEST_SUITE(suite, configs) \
static int registered_##suite = \
(grpc_core::CoreEnd2endTestRegistry::Get() \
.template RegisterSuiteT<suite>(#suite, configs), \
0)
#endif // GRPC_TEST_CORE_END2END_END2END_TESTS_H