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.

426 lines
13 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
#include "test/core/end2end/cq_verifier.h"
#include <inttypes.h>
10 years ago
#include <stdio.h>
#include <string.h>
#include <algorithm>
#include <initializer_list>
#include <limits>
#include <string>
#include <utility>
#include <vector>
#include "absl/strings/escaping.h"
#include "absl/strings/str_cat.h"
#include "absl/strings/str_format.h"
#include "absl/strings/str_join.h"
#include "absl/strings/string_view.h"
#include "gtest/gtest.h"
10 years ago
#include <grpc/byte_buffer.h>
#include <grpc/compression.h>
#include <grpc/grpc.h>
#include <grpc/slice.h>
#include <grpc/slice_buffer.h>
10 years ago
#include <grpc/support/log.h>
#include <grpc/support/time.h>
#include "src/core/lib/compression/message_compress.h"
#include "src/core/lib/gprpp/crash.h"
#include "src/core/lib/gprpp/debug_location.h"
#include "src/core/lib/gprpp/match.h"
9 years ago
#include "src/core/lib/surface/event_string.h"
#include "test/core/util/test_config.h"
10 years ago
// a set of metadata we expect to find on an event
10 years ago
typedef struct metadata {
size_t count;
size_t cap;
char** keys;
char** values;
10 years ago
} metadata;
static int has_metadata(const grpc_metadata* md, size_t count, const char* key,
const char* value) {
10 years ago
size_t i;
for (i = 0; i < count; i++) {
if (0 == grpc_slice_str_cmp(md[i].key, key) &&
0 == grpc_slice_str_cmp(md[i].value, value)) {
10 years ago
return 1;
}
}
return 0;
}
int contains_metadata(grpc_metadata_array* array, const char* key,
const char* value) {
return has_metadata(array->metadata, array->count, key, value);
}
static int has_metadata_slices(const grpc_metadata* md, size_t count,
grpc_slice key, grpc_slice value) {
size_t i;
for (i = 0; i < count; i++) {
if (grpc_slice_eq(md[i].key, key) && grpc_slice_eq(md[i].value, value)) {
return 1;
}
}
return 0;
}
int contains_metadata_slices(grpc_metadata_array* array, grpc_slice key,
grpc_slice value) {
return has_metadata_slices(array->metadata, array->count, key, value);
}
static grpc_slice merge_slices(grpc_slice* slices, size_t nslices) {
10 years ago
size_t i;
size_t len = 0;
uint8_t* cursor;
grpc_slice out;
10 years ago
for (i = 0; i < nslices; i++) {
len += GRPC_SLICE_LENGTH(slices[i]);
10 years ago
}
out = grpc_slice_malloc(len);
cursor = GRPC_SLICE_START_PTR(out);
10 years ago
for (i = 0; i < nslices; i++) {
8 years ago
memcpy(cursor, GRPC_SLICE_START_PTR(slices[i]),
GRPC_SLICE_LENGTH(slices[i]));
cursor += GRPC_SLICE_LENGTH(slices[i]);
10 years ago
}
return out;
}
int raw_byte_buffer_eq_slice(grpc_byte_buffer* rbb, grpc_slice b) {
grpc_slice a;
int ok;
if (!rbb) return 0;
a = merge_slices(rbb->data.raw.slice_buffer.slices,
rbb->data.raw.slice_buffer.count);
ok = GRPC_SLICE_LENGTH(a) == GRPC_SLICE_LENGTH(b) &&
0 == memcmp(GRPC_SLICE_START_PTR(a), GRPC_SLICE_START_PTR(b),
GRPC_SLICE_LENGTH(a));
if (!ok) {
gpr_log(GPR_ERROR,
"SLICE MISMATCH: left_length=%" PRIuPTR " right_length=%" PRIuPTR,
GRPC_SLICE_LENGTH(a), GRPC_SLICE_LENGTH(b));
std::string out;
const char* a_str = reinterpret_cast<const char*>(GRPC_SLICE_START_PTR(a));
const char* b_str = reinterpret_cast<const char*>(GRPC_SLICE_START_PTR(b));
for (size_t i = 0; i < std::max(GRPC_SLICE_LENGTH(a), GRPC_SLICE_LENGTH(b));
i++) {
if (i >= GRPC_SLICE_LENGTH(a)) {
absl::StrAppend(&out, "\u001b[36m", // cyan
absl::CEscape(absl::string_view(&b_str[i], 1)),
"\u001b[0m");
} else if (i >= GRPC_SLICE_LENGTH(b)) {
absl::StrAppend(&out, "\u001b[35m", // magenta
absl::CEscape(absl::string_view(&a_str[i], 1)),
"\u001b[0m");
} else if (a_str[i] == b_str[i]) {
absl::StrAppend(&out, absl::CEscape(absl::string_view(&a_str[i], 1)));
} else {
absl::StrAppend(&out, "\u001b[31m", // red
absl::CEscape(absl::string_view(&a_str[i], 1)),
"\u001b[33m", // yellow
absl::CEscape(absl::string_view(&b_str[i], 1)),
"\u001b[0m");
}
gpr_log(GPR_ERROR, "%s", out.c_str());
}
}
grpc_slice_unref(a);
grpc_slice_unref(b);
10 years ago
return ok;
}
int byte_buffer_eq_slice(grpc_byte_buffer* bb, grpc_slice b) {
if (bb == nullptr) return 0;
if (bb->data.raw.compression > GRPC_COMPRESS_NONE) {
grpc_slice_buffer decompressed_buffer;
grpc_slice_buffer_init(&decompressed_buffer);
Move compression related metadata to the new system (#28223) * Eliminate most of grpc_message metadata handling * Eliminate most of host metadata handling * Remove more callouts without fixing code * fiiixes * typo * Automated change: Fix sanity tests * try-shrink * Automated change: Fix sanity tests * size tweaks * less tricks * deunique * commonize * commonize * Automated change: Fix sanity tests * size tuning, fixes * Automated change: Fix sanity tests * fix * size tuning, fixes * remove constexpr * fix * reuse code * fix * tweak code * more tweaks * tell no lies * fixes * fixes * Automated change: Fix sanity tests * fix * fix * fix * fix * fix? * fix binder * fix * fix * fixes * Automated change: Fix sanity tests * fix * initial refactoring * optimize status encoding * Automated change: Fix sanity tests * Automated change: Fix sanity tests * content-type * Automated change: Fix sanity tests * clang-format * fix * Move colon prefixed metadata * Automated change: Fix sanity tests * Automated change: Fix sanity tests * try to fix windows failure * try and scale sizes better * ambiguity fix? * wip metadatavalueasslice * Fix status code for resource exhaustion * Revert "Revert "Move a bunch of slice typed metadata to new system (#28107)" (#28208)" This reverts commit 771758706396560d6eb91ef6058cdf0e1e65341e. * fix test * Automated change: Fix sanity tests * Automated change: Fix sanity tests * slice helper * x * noinline * try and scale sizes better * Automated change: Fix sanity tests * fixes * fix * fix * fixes * fix build * fix overflow * progress * Automated change: Fix sanity tests * fix * initial work * progress * fix * fix * Automated change: Fix sanity tests * progress * fix * fix * fix * Automated change: Fix sanity tests * fix * compressor for path/authority * Automated change: Fix sanity tests * Automated change: Fix sanity tests * legalize * legalize * status-enc * fmt * fix * fix * fix * fix * fix/opt * fix * fix * fix * fix * Automated change: Fix sanity tests * fix * comment * fmt * remove arg * Automated change: Fix sanity tests * remove name * Automated change: Fix sanity tests * add specialized encoders for compression metadata * review feedback * fix * missoing files * Small improvement in memory usage and performance * Automated change: Fix sanity tests * Automated change: Fix sanity tests * Fix caching * Automated change: Fix sanity tests * fix crash in alts * default everything * Automated change: Fix sanity tests * review feedback * fixes * fixes * fixes * Automated change: Fix sanity tests * speedup * fix * fix * fixes * fixes * fixes * Automated change: Fix sanity tests * fix * remove debug code Co-authored-by: ctiller <ctiller@users.noreply.github.com>
3 years ago
GPR_ASSERT(grpc_msg_decompress(bb->data.raw.compression,
&bb->data.raw.slice_buffer,
&decompressed_buffer));
grpc_byte_buffer* rbb = grpc_raw_byte_buffer_create(
decompressed_buffer.slices, decompressed_buffer.count);
int ret_val = raw_byte_buffer_eq_slice(rbb, b);
grpc_byte_buffer_destroy(rbb);
grpc_slice_buffer_destroy(&decompressed_buffer);
return ret_val;
}
return raw_byte_buffer_eq_slice(bb, b);
}
int byte_buffer_eq_string(grpc_byte_buffer* bb, const char* str) {
return byte_buffer_eq_slice(bb, grpc_slice_from_copied_string(str));
}
namespace {
bool IsProbablyInteger(void* p) {
return reinterpret_cast<uintptr_t>(p) < 1000000 ||
(reinterpret_cast<uintptr_t>(p) >
std::numeric_limits<uintptr_t>::max() - 10);
}
std::string TagStr(void* tag) {
if (IsProbablyInteger(tag)) {
return absl::StrFormat("tag(%" PRIdPTR ")",
reinterpret_cast<intptr_t>(tag));
} else {
return absl::StrFormat("%p", tag);
10 years ago
}
}
} // namespace
namespace grpc_core {
[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
CqVerifier::CqVerifier(
grpc_completion_queue* cq, absl::AnyInvocable<void(Failure) const> fail,
absl::AnyInvocable<
void(grpc_event_engine::experimental::EventEngine::Duration) const>
step_fn)
: cq_(cq), fail_(std::move(fail)), step_fn_(std::move(step_fn)) {}
10 years ago
CqVerifier::~CqVerifier() { Verify(); }
std::string CqVerifier::Expectation::ToString() const {
return absl::StrCat(
location.file(), ":", location.line(), ": ", TagStr(tag), " ",
Match(
result,
[](bool success) {
return absl::StrCat("success=", success ? "true" : "false");
},
[](Maybe) { return std::string("maybe"); },
[](AnyStatus) { return std::string("any success value"); },
[promises] Run C++ end to end tests with server promises (#32537) Expand server promises to run with C++ end2end tests. Across connected_channel/call/batch_builder/pipe/transport: - fix a bug where read errors weren't propagated from transport to call so that we can populate failed_before_recv_message for the c++ bindings - ensure those errors are not, however, used to populate the returned call status Add a new latch call arg to lazily propagate the bound CQ for a server call (and client call, but here it's used degenerately - it's always populated). This allows server calls to be properly bound to pollsets.(1)/(2) In call.cc: - move some profiling code from FilterStackCall to Call, and then use it in PromiseBasedCall (this should be cleaned up with tracing work) - implement GetServerAuthority In server.cc: - use an RAII pattern on `MatchResult` to avoid a bug whereby a tag could be dropped if we cancel a request after it's been matched but before it's published - fix deadline export to ServerContext In resource_quota_server.cc: - fix some long standing flakes (that were finally obvious with the new test code) - it's legal here to have client calls not arrive at the server due to resource starvation, work through that (includes adding expectations during a `Step` call, which required some small tweaks to cq_verifier) In the C++ end2end_test.cc: - strengthen a flaky test so it passes consistently (it's likely we'll revisit this with the fuzzing efforts to strengthen it into an actually robust test) (1) It's time to remove this concept (2) Surprisingly the only test that *reliably* demonstrates this not being done is time_change_test --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
[](const PerformAction&) {
return std::string("perform some action");
},
[](const MaybePerformAction&) {
return std::string("maybe perform action");
}));
}
std::string CqVerifier::Expectation::ToShortString() const {
return absl::StrCat(
TagStr(tag),
Match(
result,
[](bool success) -> std::string {
if (!success) return "-❌";
return "-✅";
},
[](Maybe) { return std::string("-❓"); },
[](AnyStatus) { return std::string("-🤷"); },
[](const PerformAction&) { return std::string("-🎬"); },
[](const MaybePerformAction&) { return std::string("-🎬❓"); }));
}
std::vector<std::string> CqVerifier::ToStrings() const {
std::vector<std::string> expectations;
expectations.reserve(expectations_.size());
for (const auto& e : expectations_) {
expectations.push_back(e.ToString());
10 years ago
}
return expectations;
}
std::string CqVerifier::ToString() const {
return absl::StrJoin(ToStrings(), "\n");
10 years ago
}
std::vector<std::string> CqVerifier::ToShortStrings() const {
std::vector<std::string> expectations;
expectations.reserve(expectations_.size());
for (const auto& e : expectations_) {
expectations.push_back(e.ToShortString());
}
return expectations;
}
std::string CqVerifier::ToShortString() const {
return absl::StrJoin(ToShortStrings(), " ");
}
void CqVerifier::FailNoEventReceived(const SourceLocation& location) const {
fail_(Failure{location, "No event received", ToStrings()});
10 years ago
}
void CqVerifier::FailUnexpectedEvent(grpc_event* ev,
const SourceLocation& location) const {
fail_(Failure{location,
absl::StrCat("Unexpected event: ", grpc_event_string(ev)),
ToStrings()});
}
namespace {
std::string CrashMessage(const CqVerifier::Failure& failure) {
std::string message =
absl::StrCat(failure.message, "\nexpectation checked @ ",
failure.location.file(), ":", failure.location.line());
if (!failure.expected.empty()) {
absl::StrAppend(&message, "\nexpected:\n");
for (const auto& line : failure.expected) {
absl::StrAppend(&message, " ", line, "\n");
}
} else {
absl::StrAppend(&message, "\nexpected nothing");
}
return message;
}
} // namespace
void CqVerifier::FailUsingGprCrashWithStdio(const Failure& failure) {
CrashWithStdio(CrashMessage(failure));
}
void CqVerifier::FailUsingGprCrash(const Failure& failure) {
Crash(CrashMessage(failure));
}
void CqVerifier::FailUsingGtestFail(const Failure& failure) {
std::string message = absl::StrCat(" ", failure.message);
if (!failure.expected.empty()) {
absl::StrAppend(&message, "\n expected:\n");
for (const auto& line : failure.expected) {
absl::StrAppend(&message, " ", line, "\n");
}
} else {
absl::StrAppend(&message, "\n expected nothing");
}
ADD_FAILURE_AT(failure.location.file(), failure.location.line()) << message;
}
namespace {
bool IsMaybe(const CqVerifier::ExpectedResult& r) {
return Match(
r, [](bool) { return false; }, [](CqVerifier::Maybe) { return true; },
[](CqVerifier::AnyStatus) { return false; },
[](const CqVerifier::PerformAction&) { return false; },
[](const CqVerifier::MaybePerformAction&) { return true; });
}
} // namespace
[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_event CqVerifier::Step(gpr_timespec deadline) {
if (step_fn_ != nullptr) {
while (true) {
grpc_event r = grpc_completion_queue_next(
cq_, gpr_inf_past(deadline.clock_type), nullptr);
if (r.type != GRPC_QUEUE_TIMEOUT) return r;
auto now = gpr_now(deadline.clock_type);
if (gpr_time_cmp(deadline, now) < 0) break;
step_fn_(Timestamp::FromTimespecRoundDown(deadline) - Timestamp::Now());
}
return grpc_event{GRPC_QUEUE_TIMEOUT, 0, nullptr};
}
return grpc_completion_queue_next(cq_, deadline, nullptr);
}
void CqVerifier::Verify(Duration timeout, SourceLocation location) {
if (expectations_.empty()) return;
if (log_verifications_) {
gpr_log(GPR_ERROR, "Verify %s for %s", ToShortString().c_str(),
timeout.ToString().c_str());
}
const gpr_timespec deadline =
grpc_timeout_milliseconds_to_deadline(timeout.millis());
while (!expectations_.empty()) {
[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_event ev = Step(deadline);
if (ev.type == GRPC_QUEUE_TIMEOUT) break;
if (ev.type != GRPC_OP_COMPLETE) {
FailUnexpectedEvent(&ev, location);
}
bool found = false;
for (auto it = expectations_.begin(); it != expectations_.end(); ++it) {
if (it->tag != ev.tag) continue;
[promises] Run C++ end to end tests with server promises (#32537) Expand server promises to run with C++ end2end tests. Across connected_channel/call/batch_builder/pipe/transport: - fix a bug where read errors weren't propagated from transport to call so that we can populate failed_before_recv_message for the c++ bindings - ensure those errors are not, however, used to populate the returned call status Add a new latch call arg to lazily propagate the bound CQ for a server call (and client call, but here it's used degenerately - it's always populated). This allows server calls to be properly bound to pollsets.(1)/(2) In call.cc: - move some profiling code from FilterStackCall to Call, and then use it in PromiseBasedCall (this should be cleaned up with tracing work) - implement GetServerAuthority In server.cc: - use an RAII pattern on `MatchResult` to avoid a bug whereby a tag could be dropped if we cancel a request after it's been matched but before it's published - fix deadline export to ServerContext In resource_quota_server.cc: - fix some long standing flakes (that were finally obvious with the new test code) - it's legal here to have client calls not arrive at the server due to resource starvation, work through that (includes adding expectations during a `Step` call, which required some small tweaks to cq_verifier) In the C++ end2end_test.cc: - strengthen a flaky test so it passes consistently (it's likely we'll revisit this with the fuzzing efforts to strengthen it into an actually robust test) (1) It's time to remove this concept (2) Surprisingly the only test that *reliably* demonstrates this not being done is time_change_test --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
auto expectation = std::move(*it);
expectations_.erase(it);
const bool expected = Match(
[promises] Run C++ end to end tests with server promises (#32537) Expand server promises to run with C++ end2end tests. Across connected_channel/call/batch_builder/pipe/transport: - fix a bug where read errors weren't propagated from transport to call so that we can populate failed_before_recv_message for the c++ bindings - ensure those errors are not, however, used to populate the returned call status Add a new latch call arg to lazily propagate the bound CQ for a server call (and client call, but here it's used degenerately - it's always populated). This allows server calls to be properly bound to pollsets.(1)/(2) In call.cc: - move some profiling code from FilterStackCall to Call, and then use it in PromiseBasedCall (this should be cleaned up with tracing work) - implement GetServerAuthority In server.cc: - use an RAII pattern on `MatchResult` to avoid a bug whereby a tag could be dropped if we cancel a request after it's been matched but before it's published - fix deadline export to ServerContext In resource_quota_server.cc: - fix some long standing flakes (that were finally obvious with the new test code) - it's legal here to have client calls not arrive at the server due to resource starvation, work through that (includes adding expectations during a `Step` call, which required some small tweaks to cq_verifier) In the C++ end2end_test.cc: - strengthen a flaky test so it passes consistently (it's likely we'll revisit this with the fuzzing efforts to strengthen it into an actually robust test) (1) It's time to remove this concept (2) Surprisingly the only test that *reliably* demonstrates this not being done is time_change_test --------- Co-authored-by: ctiller <ctiller@users.noreply.github.com>
2 years ago
expectation.result,
[ev](bool success) { return ev.success == success; },
[ev](Maybe m) {
if (m.seen != nullptr) *m.seen = true;
return ev.success != 0;
},
[ev](AnyStatus a) {
if (a.result != nullptr) *a.result = ev.success;
return true;
},
[ev](const PerformAction& action) {
action.action(ev.success);
return true;
},
[ev](const MaybePerformAction& action) {
action.action(ev.success);
return true;
});
if (!expected) {
FailUnexpectedEvent(&ev, location);
}
found = true;
break;
}
if (!found) FailUnexpectedEvent(&ev, location);
if (AllMaybes()) break;
}
expectations_.erase(
std::remove_if(expectations_.begin(), expectations_.end(),
[](const Expectation& e) { return IsMaybe(e.result); }),
expectations_.end());
if (!expectations_.empty()) FailNoEventReceived(location);
}
bool CqVerifier::AllMaybes() const {
for (const auto& e : expectations_) {
if (!IsMaybe(e.result)) return false;
10 years ago
}
return true;
10 years ago
}
void CqVerifier::VerifyEmpty(Duration timeout, SourceLocation location) {
if (log_verifications_) {
gpr_log(GPR_ERROR, "Verify empty completion queue for %s",
timeout.ToString().c_str());
}
const gpr_timespec deadline =
gpr_time_add(gpr_now(GPR_CLOCK_MONOTONIC), timeout.as_timespec());
GPR_ASSERT(expectations_.empty());
[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_event ev = Step(deadline);
if (ev.type != GRPC_QUEUE_TIMEOUT) {
FailUnexpectedEvent(&ev, location);
}
10 years ago
}
void CqVerifier::Expect(void* tag, ExpectedResult result,
SourceLocation location) {
expectations_.push_back(Expectation{location, tag, std::move(result)});
}
} // namespace grpc_core