[rocprofiler-sdk] Buffer finalization fixes and HSA ABI 0x09 support (#2318)
* [rocprofiler-sdk] Fix buffer flush ordering and sanitizer CI improvements Buffer Pool Design ------------------ Replace the fixed array-based double buffer with a dynamic pool design to fix race conditions that caused "internal correlation id was retired prematurely" errors. The original design had a race where flush callbacks could be delivered out-of-order: when buffer 0 fills and begins flushing, writes go to buffer 1. If buffer 1 fills before buffer 0's flush completes, the buffer index wraps back to 0 (which may still be flushing). Independent flush tasks submitted to the thread pool can complete out of order. The new pool design: - Uses a std::deque of buffer instances that grows as needed - Allocates buffers from the pool when the current buffer needs to flush - Serializes flushes with a mutex to ensure FIFO callback ordering - Returns buffers to the pool after flush completion - Eliminates the race between buffer selection and write operations New Unit Tests -------------- - buffer_correlation_ordering.cpp: Tests that API records are always delivered before their corresponding retirement records - buffer_ordering_stress.cpp: Stress tests buffer flush ordering under high contention with multiple threads rapidly filling buffers HSA Tool Hooks -------------- Added hsa_tool_hooks.cpp/hpp to register an HSA OnUnload callback that waits for pending flush tasks before tool finalization, preventing "retired prematurely" errors during HSA shutdown. Sanitizer Improvements ---------------------- - LSAN: Set fast_unwind_on_malloc=1 to prevent deadlock in libgcc unwinder - LSAN: Added suppressions for external tools (liblzma, liblsan, seq, strdup) - TSAN: Added suppression for false positive on C++11 thread-safe static initialization in create_write_functor - ASAN/UBSAN: Added patterns for known issues in HSA runtime, HIP, perfetto - Disabled attachment tests for sanitizers due to library preloading issues Other Fixes ----------- - Thread-trace agent test: Use heap-allocated callback state - Correlation ID: Refactored reference counting and finalization ordering * [rocprofiler-sdk] Revert buffer pool design changes Revert buffer.cpp and buffer.hpp to the original double-buffer design from develop branch. The pool-based redesign introduced concerns about: - Signal safety (mutex vs atomic_flag) - API changes (flush() return type) - Complexity of the new design This revert removes: - Dynamic buffer pool with std::deque - std::mutex/condition_variable synchronization - buffer_correlation_ordering.cpp test - buffer_ordering_stress.cpp test The underlying buffer flush ordering issue will need to be addressed with a different approach that preserves the original API and synchronization characteristics. * [rocprofiler-sdk] Consistent fini_status checks to prevent correlation ID creation during finalization - Revert TOCTOU CAS loop change in sub_ref_count() - not needed with consistent checks - Add fini_status check in correlation_tracing_service::construct() with ROCP_CI_LOG warning - Add nullptr checks at all construct() call sites (queue.cpp, async_copy.cpp, memory_allocation.cpp) - Change all 'get_fini_status() > 0' to '!= 0' for consistent behavior: - hsa/queue.cpp (lines 105, 210) - hsa/async_copy.cpp (line 344) - hsa/hsa_barrier.cpp (line 43) - buffer.cpp (lines 107, 138, 185) This ensures no correlation IDs are created once finalization starts (fini_status != 0), preventing races between finalization and ongoing tracing operations. * [rocprofiler-sdk] Replace arrival-order checks with timestamp-based temporal validation Buffer records are not guaranteed to arrive in any specific order. Tests and samples should use timestamps for temporal ordering validation instead. Changes: - samples/external_correlation_id_request: Replace 'retired prematurely' arrival order check with timestamp-based validation that retirement timestamp >= max(end_timestamps) for records with the same correlation ID - tests/external_correlation.cpp: Remove EXPECT_GT(corr_id, last_corr_id) check - tests/registration.cpp: Remove EXPECT_GT(corr_id, last_corr_id) check - tests/roctx.cpp: Remove EXPECT_GT(corr_id, last_corr_id) check Correlation IDs are not guaranteed to be monotonically increasing when records are sorted by timestamp. Temporal ordering should be validated using the timestamp fields in each record. * [rocprofiler-sdk] Revert external/CMakeLists.txt SYSTEM keyword removal Restore the SYSTEM keyword to target_include_directories for rocprofiler-sdk-fmt to match develop branch. * [rccl] Remove orphaned rocSHMEM gitlink Remove orphaned submodule reference that was introduced during a merge but never had a corresponding .gitmodules entry, causing CI failures with "fatal: no submodule mapping found in .gitmodules". * [rocprofiler-sdk] Add HSA ABI version 0x09 support Add ABI checks for HSA_AMD_EXT_API_TABLE_STEP_VERSION 0x09 which introduces hsa_amd_counted_queue_acquire and hsa_amd_counted_queue_release functions (added in rocr-runtime SWDEV-561708). * [rocprofiler-sdk] Handle finalized status gracefully in buffer flush operations This commit consolidates fixes for handling the finalization status during buffer flush operations across the SDK. Changes: - Tool and samples: Handle ROCPROFILER_STATUS_ERROR_FINALIZED gracefully when flushing buffers, as this indicates buffers were already flushed during finalization (not an error condition) - HSA handlers (queue.cpp, async_copy.cpp, hsa_barrier.cpp): Use > 0 check for fini_status to allow operations during finalization process - buffer.cpp: Revert fini_status checks to use > 0 for consistency - correlation_id.cpp: Add fini_status > 0 check with ROCP_TRACE logging to prevent correlation ID creation after finalization starts Files modified: - source/lib/rocprofiler-sdk-tool/tool.cpp - tests/tools/json-tool.cpp - source/lib/rocprofiler-sdk/tests/registration.cpp - source/lib/rocprofiler-sdk/tests/roctx.cpp - samples/api_buffered_tracing/client.cpp - samples/counter_collection/buffered_client.cpp - samples/counter_collection/device_counting_async_client.cpp - samples/external_correlation_id_request/client.cpp - samples/pc_sampling/client.cpp - source/lib/rocprofiler-sdk/buffer.cpp - source/lib/rocprofiler-sdk/context/correlation_id.cpp - source/lib/rocprofiler-sdk/hsa/queue.cpp - source/lib/rocprofiler-sdk/hsa/async_copy.cpp - source/lib/rocprofiler-sdk/hsa/hsa_barrier.cpp * [rocprofiler-sdk] Remove hsa_tool_hooks and simplify buffer flush handling Remove the hsa_tool_hooks infrastructure and simplify buffer flush calls in samples and tools. The ERROR_FINALIZED handling was overly complex and the hsa_tool_hooks OnUnload synchronization is no longer needed. Changes: - Remove hsa_tool_hooks.cpp/hpp and related registration.cpp code - Simplify buffer flush calls in samples to use direct ROCPROFILER_CALL - Simplify buffer flush in tool.cpp and json-tool.cpp - Remove ERROR_FINALIZED special handling from test files Co-Authored-By: Claude <noreply@anthropic.com> * [rocprofiler-sdk] Fix output_stream move semantics to null source pointers The default move constructor and move assignment operator for output_stream did not null out the source's pointers after the move. This caused double-close when the moved-from temporary was destroyed, leading to use-after-free crashes (SIGSEGV in std::ostream::sentry). Co-Authored-By: Claude <noreply@anthropic.com> * [rocprofiler-sdk] Improve Perfetto trace writer and sanitizer configuration - generatePerfetto.cpp: Move output_stream into shared_state to prevent use-after-free race conditions during Perfetto callback execution - run-ci.py: Simplify and consolidate sanitizer environment variable configuration for better maintainability Co-Authored-By: Claude <noreply@anthropic.com> * [rocprofiler-sdk] Revert run-ci.py changes that broke sanitizer suppressions The previous changes removed MEMCHECK_SANITIZER_OPTIONS which is required for CTest to properly pass suppression files to the sanitizers during memcheck runs. Co-Authored-By: Claude <noreply@anthropic.com> * Revert "[rccl] Remove orphaned rocSHMEM gitlink" This reverts commit 1ad21003941355658fff8114fa27768f11a948f7. * [rocprofiler-sdk] Revert registration.cpp changes Revert changes to registration.cpp to match develop branch. Co-Authored-By: Claude <noreply@anthropic.com> * [rocprofiler-sdk] Remove suppression file content printing from run-ci.py Co-Authored-By: Claude <noreply@anthropic.com> * Fix output_stream move ctor/assignment operator * Fix erroneous revert of registration.cpp * Fix handling of fini status in correlation ID construction * [rocprofiler-sdk] Fix OMPT segfault during finalization Add nullptr checks in OMPT tracing code to handle the case where correlation_tracing_service::construct() returns nullptr during finalization. This fixes segfaults in openmp-target-sample and tests.integration.execute.openmp-tools. The correlation ID construction now returns nullptr when fini_status > 0, but the OMPT callbacks were not checking for this, causing crashes when dereferencing the null pointer during OpenMP runtime shutdown. Changes: - event_common(): Return nullptr early if correlation ID is null - event(): Check for nullptr before calling sub_ref_count() - ompt_task_create_callback(): Return early if correlation ID is null - ompt_task_schedule_callback(): Return early if correlation ID is null * [rocprofiler-sdk] Fix HSA API tracing segfault during finalization Add nullptr check in hsa_api_impl::functor after correlation ID construction. During finalization, correlation_service::construct() returns nullptr, and without this check the code would dereference the null pointer when accessing corr_id->internal. This fixes the SEGV at address 0x000000000008 (null + 8 byte offset) that occurs when HSA async event threads call hsa_signal_destroy during runtime shutdown after finalization has started. --------- Co-authored-by: Claude <noreply@anthropic.com> Co-authored-by: Jonathan R. Madsen <jonathanrmadsen@gmail.com>
This commit is contained in:
Benjamin Welton
GitHub
@@ -360,6 +360,10 @@ ROCPROFILER_ENUM_LABEL(ROCPROFILER_HSA_AMD_EXT_API_ID_hsa_amd_portable_export_dm
|
||||
ROCPROFILER_ENUM_LABEL(ROCPROFILER_HSA_AMD_EXT_API_ID_hsa_amd_ais_file_write);
|
||||
ROCPROFILER_ENUM_LABEL(ROCPROFILER_HSA_AMD_EXT_API_ID_hsa_amd_ais_file_read);
|
||||
# endif
|
||||
# if HSA_AMD_EXT_API_TABLE_STEP_VERSION >= 0x09
|
||||
ROCPROFILER_ENUM_LABEL(ROCPROFILER_HSA_AMD_EXT_API_ID_hsa_amd_counted_queue_acquire);
|
||||
ROCPROFILER_ENUM_LABEL(ROCPROFILER_HSA_AMD_EXT_API_ID_hsa_amd_counted_queue_release);
|
||||
# endif
|
||||
#endif
|
||||
|
||||
#if HSA_AMD_EXT_API_TABLE_MAJOR_VERSION == 0x01
|
||||
@@ -383,6 +387,8 @@ static_assert(ROCPROFILER_HSA_AMD_EXT_API_ID_LAST == 73);
|
||||
static_assert(ROCPROFILER_HSA_AMD_EXT_API_ID_LAST == 74);
|
||||
# elif HSA_AMD_EXT_API_TABLE_STEP_VERSION == 0x08
|
||||
static_assert(ROCPROFILER_HSA_AMD_EXT_API_ID_LAST == 76);
|
||||
# elif HSA_AMD_EXT_API_TABLE_STEP_VERSION == 0x09
|
||||
static_assert(ROCPROFILER_HSA_AMD_EXT_API_ID_LAST == 78);
|
||||
# else
|
||||
# if !defined(ROCPROFILER_UNSAFE_NO_VERSION_CHECK) && \
|
||||
(defined(ROCPROFILER_CI) && ROCPROFILER_CI > 0)
|
||||
|
||||
@@ -125,6 +125,10 @@ typedef enum rocprofiler_hsa_amd_ext_api_id_t // NOLINT(performance-enum-size)
|
||||
ROCPROFILER_HSA_AMD_EXT_API_ID_hsa_amd_ais_file_write,
|
||||
ROCPROFILER_HSA_AMD_EXT_API_ID_hsa_amd_ais_file_read,
|
||||
# endif
|
||||
# if HSA_AMD_EXT_API_TABLE_STEP_VERSION >= 0x09
|
||||
ROCPROFILER_HSA_AMD_EXT_API_ID_hsa_amd_counted_queue_acquire,
|
||||
ROCPROFILER_HSA_AMD_EXT_API_ID_hsa_amd_counted_queue_release,
|
||||
# endif
|
||||
#endif
|
||||
|
||||
ROCPROFILER_HSA_AMD_EXT_API_ID_LAST,
|
||||
|
||||
@@ -1464,6 +1464,22 @@ typedef union rocprofiler_hsa_api_args_t
|
||||
int32_t* status;
|
||||
} hsa_amd_ais_file_read;
|
||||
# endif
|
||||
# if HSA_AMD_EXT_API_TABLE_STEP_VERSION >= 0x09
|
||||
struct
|
||||
{
|
||||
hsa_agent_t agent;
|
||||
hsa_queue_type_t type;
|
||||
hsa_amd_queue_priority_t priority;
|
||||
void (*callback)(hsa_status_t status, hsa_queue_t* source, void* data);
|
||||
void* data;
|
||||
uint64_t flags;
|
||||
hsa_queue_t** queue;
|
||||
} hsa_amd_counted_queue_acquire;
|
||||
struct
|
||||
{
|
||||
hsa_queue_t* queue;
|
||||
} hsa_amd_counted_queue_release;
|
||||
# endif
|
||||
#endif
|
||||
} rocprofiler_hsa_api_args_t;
|
||||
|
||||
|
||||
@@ -98,5 +98,15 @@ add_string_entry(std::string_view name)
|
||||
|
||||
return _hash_v;
|
||||
}
|
||||
|
||||
// Clear string entry cache for attach/detach cycles
|
||||
void
|
||||
clear_string_entries()
|
||||
{
|
||||
if(!get_string_array()) return;
|
||||
|
||||
auto _lk = std::unique_lock<std::shared_mutex>{get_sync()};
|
||||
get_string_array()->clear();
|
||||
}
|
||||
} // namespace common
|
||||
} // namespace rocprofiler
|
||||
|
||||
@@ -38,5 +38,9 @@ get_string_entry(size_t hash);
|
||||
|
||||
size_t
|
||||
add_string_entry(std::string_view name);
|
||||
|
||||
// Clear string entry cache (for attach/detach)
|
||||
void
|
||||
clear_string_entries();
|
||||
} // namespace common
|
||||
} // namespace rocprofiler
|
||||
|
||||
@@ -39,6 +39,7 @@
|
||||
#include <future>
|
||||
#include <iostream>
|
||||
#include <map>
|
||||
#include <memory>
|
||||
#include <thread>
|
||||
#include <unordered_map>
|
||||
#include <utility>
|
||||
@@ -1204,44 +1205,46 @@ write_perfetto(
|
||||
tracing_session->FlushBlocking();
|
||||
tracing_session->StopBlocking();
|
||||
|
||||
auto filename = std::string{"results"};
|
||||
auto ofs = get_output_stream(ocfg, filename, ".pftrace");
|
||||
|
||||
auto amount_read = std::atomic<size_t>{0};
|
||||
auto is_done = std::promise<void>{};
|
||||
auto _mtx = std::mutex{};
|
||||
auto _reader = [&ofs, &_mtx, &is_done, &amount_read](
|
||||
::perfetto::TracingSession::ReadTraceCallbackArgs _args) {
|
||||
auto _lk = std::unique_lock<std::mutex>{_mtx};
|
||||
if(_args.data && _args.size > 0)
|
||||
{
|
||||
ROCP_TRACE << "Writing " << _args.size << " B to trace...";
|
||||
// Write the trace data into file
|
||||
ofs.stream->write(_args.data, _args.size);
|
||||
amount_read += _args.size;
|
||||
}
|
||||
ROCP_INFO_IF(!_args.has_more && amount_read > 0)
|
||||
<< "Wrote " << amount_read << " B to perfetto trace file";
|
||||
if(!_args.has_more) is_done.set_value();
|
||||
struct read_trace_state
|
||||
{
|
||||
std::mutex mtx{};
|
||||
std::atomic<size_t> amount_read{0};
|
||||
output_stream ofs{};
|
||||
};
|
||||
|
||||
auto state = std::make_shared<read_trace_state>();
|
||||
state->ofs = get_output_stream(ocfg, std::string{"results"}, ".pftrace");
|
||||
|
||||
for(size_t i = 0; i < 2; ++i)
|
||||
{
|
||||
ROCP_TRACE << "Reading trace...";
|
||||
amount_read = 0;
|
||||
is_done = std::promise<void>{};
|
||||
|
||||
auto is_done = std::make_shared<std::promise<void>>();
|
||||
auto _reader = [state, is_done](::perfetto::TracingSession::ReadTraceCallbackArgs _args) {
|
||||
auto _lk = std::unique_lock<std::mutex>{state->mtx};
|
||||
if(_args.data && _args.size > 0)
|
||||
{
|
||||
ROCP_TRACE << "Writing " << _args.size << " B to trace...";
|
||||
// Write the trace data into file
|
||||
state->ofs.stream->write(_args.data, _args.size);
|
||||
state->amount_read += _args.size;
|
||||
}
|
||||
ROCP_INFO_IF(!_args.has_more && state->amount_read > 0)
|
||||
<< "Wrote " << state->amount_read << " B to perfetto trace file";
|
||||
if(!_args.has_more) is_done->set_value();
|
||||
};
|
||||
tracing_session->ReadTrace(_reader);
|
||||
is_done.get_future().wait();
|
||||
is_done->get_future().wait();
|
||||
}
|
||||
|
||||
ROCP_TRACE << "Destroying tracing session...";
|
||||
tracing_session.reset();
|
||||
|
||||
ROCP_TRACE << "Flushing trace output stream...";
|
||||
(*ofs.stream) << std::flush;
|
||||
(*state->ofs.stream) << std::flush;
|
||||
|
||||
ROCP_TRACE << "Destroying trace output stream...";
|
||||
ofs.close();
|
||||
state->ofs.close();
|
||||
}
|
||||
|
||||
} // namespace tool
|
||||
|
||||
@@ -1480,6 +1480,10 @@ write_rocpd(
|
||||
}
|
||||
|
||||
SQLITE3_CHECK(sqlite3_close_v2(conn));
|
||||
|
||||
// Clear UUID/GUID state at end of write to prepare for potential re-attach
|
||||
get_guid().clear();
|
||||
get_uuid().clear();
|
||||
}
|
||||
} // namespace tool
|
||||
} // namespace rocprofiler
|
||||
|
||||
@@ -51,10 +51,24 @@ struct output_stream
|
||||
{}
|
||||
|
||||
~output_stream() { close(); }
|
||||
output_stream(const output_stream&) = delete;
|
||||
output_stream(output_stream&&) noexcept = default;
|
||||
output_stream(const output_stream&) = delete;
|
||||
output_stream& operator=(const output_stream&) = delete;
|
||||
output_stream& operator=(output_stream&&) noexcept = default;
|
||||
|
||||
output_stream(output_stream&& other) noexcept
|
||||
{
|
||||
std::swap(stream, other.stream);
|
||||
std::swap(dtor, other.dtor);
|
||||
}
|
||||
|
||||
output_stream& operator=(output_stream&& other) noexcept
|
||||
{
|
||||
if(this != &other)
|
||||
{
|
||||
std::swap(stream, other.stream);
|
||||
std::swap(dtor, other.dtor);
|
||||
}
|
||||
return *this;
|
||||
}
|
||||
|
||||
explicit operator bool() const { return stream != nullptr; }
|
||||
|
||||
|
||||
@@ -35,7 +35,6 @@ except Exception:
|
||||
|
||||
from . import libpyrocpd
|
||||
|
||||
|
||||
__all__ = ["format_path", "output_config", "add_args", "add_generic_args"]
|
||||
|
||||
|
||||
|
||||
@@ -127,29 +127,29 @@ PerfettoSession::~PerfettoSession()
|
||||
auto filename = std::string{"results"};
|
||||
auto ofs = tool::get_output_stream(config, filename, ".pftrace", std::ios::binary);
|
||||
|
||||
auto amount_read = std::atomic<size_t>{0};
|
||||
auto is_done = std::promise<void>{};
|
||||
auto _mtx = std::mutex{};
|
||||
auto _reader = [&ofs, &_mtx, &is_done, &amount_read](
|
||||
::perfetto::TracingSession::ReadTraceCallbackArgs _args) {
|
||||
auto _lk = std::unique_lock<std::mutex>{_mtx};
|
||||
if(_args.data && _args.size > 0)
|
||||
{
|
||||
ROCP_TRACE << "Writing " << _args.size << " B to trace...";
|
||||
// Write the trace data into file
|
||||
ofs.stream->write(_args.data, _args.size);
|
||||
amount_read += _args.size;
|
||||
}
|
||||
ROCP_INFO_IF(!_args.has_more && amount_read > 0)
|
||||
<< "Wrote " << amount_read << " B to perfetto trace file";
|
||||
if(!_args.has_more) is_done.set_value();
|
||||
};
|
||||
|
||||
// NOTE: These variables must be inside the loop to avoid a TSAN race condition.
|
||||
// If is_done is declared outside and reassigned each iteration, the promise's internal
|
||||
// mutex can be destroyed while the callback thread is still unlocking it after set_value().
|
||||
for(size_t i = 0; i < 2; ++i)
|
||||
{
|
||||
ROCP_TRACE << "Reading trace...";
|
||||
amount_read = 0;
|
||||
is_done = std::promise<void>{};
|
||||
auto amount_read = std::atomic<size_t>{0};
|
||||
auto is_done = std::promise<void>{};
|
||||
auto _mtx = std::mutex{};
|
||||
auto _reader = [&ofs, &_mtx, &is_done, &amount_read](
|
||||
::perfetto::TracingSession::ReadTraceCallbackArgs _args) {
|
||||
auto _lk = std::unique_lock<std::mutex>{_mtx};
|
||||
if(_args.data && _args.size > 0)
|
||||
{
|
||||
ROCP_TRACE << "Writing " << _args.size << " B to trace...";
|
||||
// Write the trace data into file
|
||||
ofs.stream->write(_args.data, _args.size);
|
||||
amount_read += _args.size;
|
||||
}
|
||||
ROCP_INFO_IF(!_args.has_more && amount_read > 0)
|
||||
<< "Wrote " << amount_read << " B to perfetto trace file";
|
||||
if(!_args.has_more) is_done.set_value();
|
||||
};
|
||||
tracing_session->ReadTrace(_reader);
|
||||
is_done.get_future().wait();
|
||||
}
|
||||
|
||||
@@ -86,28 +86,6 @@ v1_get_query_info(hsa_agent_t agent, const counters::Metric& metric)
|
||||
return query;
|
||||
}
|
||||
|
||||
uint32_t
|
||||
v1_get_block_counters(hsa_agent_t agent, const hsa_ven_amd_aqlprofile_event_t& event)
|
||||
{
|
||||
hsa_ven_amd_aqlprofile_profile_t query = {.agent = agent,
|
||||
.type = HSA_VEN_AMD_AQLPROFILE_EVENT_TYPE_PMC,
|
||||
.events = &event,
|
||||
.event_count = 1,
|
||||
.parameters = nullptr,
|
||||
.parameter_count = 0,
|
||||
.output_buffer = {nullptr, 0},
|
||||
.command_buffer = {nullptr, 0}};
|
||||
uint32_t max_block_counters = 0;
|
||||
if(hsa_ven_amd_aqlprofile_get_info(&query,
|
||||
HSA_VEN_AMD_AQLPROFILE_INFO_BLOCK_COUNTERS,
|
||||
&max_block_counters) != HSA_STATUS_SUCCESS)
|
||||
{
|
||||
throw std::runtime_error(fmt::format("AQL failed to max block info for counter {}",
|
||||
static_cast<int64_t>(event.block_name)));
|
||||
}
|
||||
return max_block_counters;
|
||||
}
|
||||
|
||||
void
|
||||
test_init()
|
||||
{
|
||||
@@ -199,44 +177,6 @@ TEST(aql_helpers, get_block_counters)
|
||||
}
|
||||
}
|
||||
|
||||
TEST(aql_helpers, get_block_counters_compare_v1)
|
||||
{
|
||||
ASSERT_EQ(hsa_init(), HSA_STATUS_SUCCESS);
|
||||
test_init();
|
||||
auto agents = agent::get_agents();
|
||||
|
||||
ASSERT_FALSE(agents.empty());
|
||||
|
||||
for(auto agent : agents)
|
||||
{
|
||||
if(agent->type == ROCPROFILER_AGENT_TYPE_CPU) continue;
|
||||
auto metrics = findDeviceMetrics(*agent, {});
|
||||
ASSERT_FALSE(metrics.empty());
|
||||
|
||||
for(auto& metric : metrics)
|
||||
{
|
||||
auto query = aql::get_query_info(agent->id, metric);
|
||||
for(unsigned block_index = 0; block_index < query.instance_count; ++block_index)
|
||||
{
|
||||
aqlprofile_pmc_event_t event = {
|
||||
.block_index = block_index,
|
||||
.event_id = static_cast<uint32_t>(std::atoi(metric.event().c_str())),
|
||||
.flags = aqlprofile_pmc_event_flags_t{0},
|
||||
.block_name = static_cast<hsa_ven_amd_aqlprofile_block_name_t>(query.id)};
|
||||
|
||||
hsa_ven_amd_aqlprofile_event_t event_v1 = {
|
||||
.block_name = static_cast<hsa_ven_amd_aqlprofile_block_name_t>(query.id),
|
||||
.block_index = block_index,
|
||||
.counter_id = static_cast<uint32_t>(std::atoi(metric.event().c_str()))};
|
||||
EXPECT_EQ(aql::get_block_counters(agent->id, event),
|
||||
v1_get_block_counters(agent::get_agent_cache(agent)->get_hsa_agent(),
|
||||
event_v1));
|
||||
}
|
||||
}
|
||||
}
|
||||
hsa_shut_down();
|
||||
}
|
||||
|
||||
TEST(aql_helpers, get_dim_info)
|
||||
{
|
||||
auto agents = agent::get_agents();
|
||||
|
||||
@@ -137,6 +137,16 @@ correlation_id::sub_kern_count()
|
||||
correlation_id*
|
||||
correlation_tracing_service::construct(uint32_t _init_ref_count)
|
||||
{
|
||||
// During finalization, we cannot create correlation IDs.
|
||||
// This can happen when HSA async threads call APIs after finalization starts.
|
||||
// This is not an error condition - just return nullptr and let the caller handle it.
|
||||
if(registration::get_fini_status() != 0)
|
||||
{
|
||||
ROCP_TRACE << "correlation_tracing_service::construct called during finalization"
|
||||
<< " (fini_status=" << registration::get_fini_status() << "), returning nullptr";
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
ROCP_FATAL_IF(_init_ref_count == 0) << "must have reference count > 0";
|
||||
|
||||
auto _internal_id = get_unique_internal_id();
|
||||
|
||||
@@ -59,6 +59,8 @@ ROCP_SDK_ENFORCE_ABI_VERSIONING(::AmdExtTable, 74);
|
||||
ROCP_SDK_ENFORCE_ABI_VERSIONING(::AmdExtTable, 75);
|
||||
#elif HSA_AMD_EXT_API_TABLE_STEP_VERSION == 0x08
|
||||
ROCP_SDK_ENFORCE_ABI_VERSIONING(::AmdExtTable, 77);
|
||||
#elif HSA_AMD_EXT_API_TABLE_STEP_VERSION == 0x09
|
||||
ROCP_SDK_ENFORCE_ABI_VERSIONING(::AmdExtTable, 79);
|
||||
#else
|
||||
INTERNAL_CI_ROCP_SDK_ENFORCE_ABI_VERSIONING(::AmdExtTable, 0);
|
||||
#endif
|
||||
@@ -299,6 +301,10 @@ ROCP_SDK_ENFORCE_ABI(::AmdExtTable, hsa_amd_portable_export_dmabuf_v2_fn, 74);
|
||||
ROCP_SDK_ENFORCE_ABI(::AmdExtTable, hsa_amd_ais_file_write_fn, 75);
|
||||
ROCP_SDK_ENFORCE_ABI(::AmdExtTable, hsa_amd_ais_file_read_fn, 76);
|
||||
#endif
|
||||
#if HSA_AMD_EXT_API_TABLE_STEP_VERSION >= 0x09
|
||||
ROCP_SDK_ENFORCE_ABI(::AmdExtTable, hsa_amd_counted_queue_acquire_fn, 77);
|
||||
ROCP_SDK_ENFORCE_ABI(::AmdExtTable, hsa_amd_counted_queue_release_fn, 78);
|
||||
#endif
|
||||
|
||||
ROCP_SDK_ENFORCE_ABI(::ImageExtTable, hsa_ext_image_get_capability_fn, 1);
|
||||
ROCP_SDK_ENFORCE_ABI(::ImageExtTable, hsa_ext_image_data_get_info_fn, 2);
|
||||
|
||||
@@ -693,6 +693,16 @@ async_copy_impl(Args... args)
|
||||
_corr_id_pop = _data->correlation_id;
|
||||
}
|
||||
|
||||
if(!_data->correlation_id)
|
||||
{
|
||||
// During finalization - cleanup and execute without tracing
|
||||
ROCP_HSA_TABLE_CALL(ERROR, get_core_table()->hsa_signal_destroy_fn(_data->rocp_signal));
|
||||
delete _data;
|
||||
return invoke(get_next_dispatch<TableIdx, OpIdx>(),
|
||||
std::move(_tied_args),
|
||||
std::make_index_sequence<N>{});
|
||||
}
|
||||
|
||||
// increase the reference count to denote that this correlation id is being used in a kernel
|
||||
_data->correlation_id->add_ref_count();
|
||||
|
||||
|
||||
@@ -336,11 +336,22 @@ hsa_api_impl<TableIdx, OpIdx>::functor(Args... args)
|
||||
return;
|
||||
}
|
||||
|
||||
auto buffer_record = common::init_public_api_struct(buffer_hsa_api_record_t{});
|
||||
auto tracer_data = common::init_public_api_struct(callback_hsa_api_data_t{});
|
||||
auto* corr_id = tracing::correlation_service::construct(ref_count);
|
||||
auto internal_corr_id = corr_id->internal;
|
||||
auto ancestor_corr_id = corr_id->ancestor;
|
||||
auto buffer_record = common::init_public_api_struct(buffer_hsa_api_record_t{});
|
||||
auto tracer_data = common::init_public_api_struct(callback_hsa_api_data_t{});
|
||||
auto* corr_id = tracing::correlation_service::construct(ref_count);
|
||||
|
||||
// During finalization, correlation ID construction may return nullptr
|
||||
if(!corr_id)
|
||||
{
|
||||
[[maybe_unused]] auto _ret = exec(info_type::get_table_func(), std::forward<Args>(args)...);
|
||||
if constexpr(!std::is_void<RetT>::value)
|
||||
return _ret;
|
||||
else
|
||||
return;
|
||||
}
|
||||
|
||||
auto internal_corr_id = corr_id->internal;
|
||||
auto ancestor_corr_id = corr_id->ancestor;
|
||||
tracing::populate_external_correlation_ids(external_corr_ids,
|
||||
thr_id,
|
||||
external_corr_id_domain_idx,
|
||||
|
||||
@@ -518,6 +518,24 @@ HSA_API_INFO_DEFINITION_V(ROCPROFILER_HSA_TABLE_ID_AmdExt,
|
||||
size_copied,
|
||||
status)
|
||||
# endif
|
||||
# if HSA_AMD_EXT_API_TABLE_STEP_VERSION >= 0x09
|
||||
HSA_API_INFO_DEFINITION_V(ROCPROFILER_HSA_TABLE_ID_AmdExt,
|
||||
ROCPROFILER_HSA_AMD_EXT_API_ID_hsa_amd_counted_queue_acquire,
|
||||
hsa_amd_counted_queue_acquire,
|
||||
hsa_amd_counted_queue_acquire_fn,
|
||||
agent,
|
||||
type,
|
||||
priority,
|
||||
callback,
|
||||
data,
|
||||
flags,
|
||||
queue)
|
||||
HSA_API_INFO_DEFINITION_V(ROCPROFILER_HSA_TABLE_ID_AmdExt,
|
||||
ROCPROFILER_HSA_AMD_EXT_API_ID_hsa_amd_counted_queue_release,
|
||||
hsa_amd_counted_queue_release,
|
||||
hsa_amd_counted_queue_release_fn,
|
||||
queue)
|
||||
# endif
|
||||
# endif
|
||||
|
||||
#elif defined(ROCPROFILER_LIB_ROCPROFILER_HSA_ASYNC_COPY_CPP_IMPL) && \
|
||||
|
||||
@@ -503,6 +503,14 @@ memory_allocation_impl(Args... args)
|
||||
_data.correlation_id->add_ref_count();
|
||||
}
|
||||
|
||||
if(!_data.correlation_id)
|
||||
{
|
||||
// During finalization - execute without tracing
|
||||
return invoke(get_next_dispatch<TableIdx, OpIdx>(),
|
||||
std::move(_tied_args),
|
||||
std::make_index_sequence<N>{});
|
||||
}
|
||||
|
||||
auto thr_id = _data.correlation_id->thread_idx;
|
||||
tracing::populate_external_correlation_ids(
|
||||
tracing_data.external_correlation_ids,
|
||||
@@ -629,6 +637,14 @@ memory_free_impl(Args... args)
|
||||
_data.correlation_id->add_ref_count();
|
||||
}
|
||||
|
||||
if(!_data.correlation_id)
|
||||
{
|
||||
// During finalization - execute without tracing
|
||||
return invoke(get_next_dispatch<TableIdx, OpIdx>(),
|
||||
std::move(_tied_args),
|
||||
std::make_index_sequence<N>{});
|
||||
}
|
||||
|
||||
auto thr_id = _data.correlation_id->thread_idx;
|
||||
tracing::populate_external_correlation_ids(
|
||||
tracing_data.external_correlation_ids,
|
||||
|
||||
@@ -70,14 +70,6 @@ namespace hsa
|
||||
{
|
||||
namespace
|
||||
{
|
||||
constexpr int64_t NUM_SIGNALS = 16;
|
||||
std::atomic<int64_t>&
|
||||
get_balanced_signal_slots()
|
||||
{
|
||||
static auto*& atomic = common::static_object<std::atomic<int64_t>>::construct(NUM_SIGNALS);
|
||||
return *atomic;
|
||||
}
|
||||
|
||||
template <typename DomainT, typename... Args>
|
||||
inline bool
|
||||
context_filter(const context::context* ctx, DomainT domain, Args... args)
|
||||
@@ -117,8 +109,6 @@ AsyncSignalHandler(hsa_signal_value_t /*signal_v*/, void* data)
|
||||
return false;
|
||||
}
|
||||
|
||||
get_balanced_signal_slots().fetch_add(1);
|
||||
|
||||
auto& shared_ptr_info = *static_cast<std::shared_ptr<Queue::queue_info_session_t>*>(data);
|
||||
auto& queue_info_session = *shared_ptr_info;
|
||||
|
||||
@@ -290,6 +280,13 @@ WriteInterceptor(const void* packets,
|
||||
_corr_id_pop = corr_id;
|
||||
}
|
||||
|
||||
if(!corr_id)
|
||||
{
|
||||
// During finalization - just write packet through without tracing
|
||||
transformed_packets.emplace_back(packets_arr[i]);
|
||||
continue;
|
||||
}
|
||||
|
||||
// increase the reference count to denote that this correlation id is being used in a kernel
|
||||
corr_id->add_ref_count();
|
||||
corr_id->add_kern_count();
|
||||
@@ -376,13 +373,6 @@ WriteInterceptor(const void* packets,
|
||||
thr_id,
|
||||
ROCPROFILER_EXTERNAL_CORRELATION_REQUEST_KERNEL_DISPATCH);
|
||||
|
||||
// If there is a lot of contention for HSA signals, then schedule out the thread
|
||||
if(get_balanced_signal_slots().fetch_sub(1) <= 0)
|
||||
{
|
||||
sched_yield();
|
||||
std::this_thread::sleep_for(std::chrono::nanoseconds(100));
|
||||
}
|
||||
|
||||
// Stores the instrumentation pkt (i.e. AQL packets for counter collection)
|
||||
// along with an ID of the client we got the packet from (this will be returned via
|
||||
// completed_cb_t)
|
||||
@@ -689,11 +679,6 @@ Queue::sync() const
|
||||
_core_api.hsa_signal_wait_relaxed_fn(
|
||||
_active_kernels, HSA_SIGNAL_CONDITION_EQ, 0, UINT64_MAX, HSA_WAIT_STATE_ACTIVE);
|
||||
}
|
||||
// get_balanced_signal_slots() increments upon kernel dispatch completion and decrements in
|
||||
// WriteInterceptor with a starting value of NUM_SIGNALS, so the get_balanced_signal_slots()
|
||||
// should be equivalent to NUM_SIGNALS if all kernel dispatches are completed
|
||||
ROCP_CI_LOG_IF(WARNING, get_balanced_signal_slots().load() != NUM_SIGNALS) << fmt::format(
|
||||
"There are {} incomplete dispatches", NUM_SIGNALS - get_balanced_signal_slots().load());
|
||||
}
|
||||
|
||||
void
|
||||
|
||||
@@ -147,6 +147,7 @@ ompt_task_create_callback(ompt_data_t* encountering_task_data,
|
||||
flags,
|
||||
has_dependences,
|
||||
codeptr_ra);
|
||||
if(!corr_id) return; // During finalization
|
||||
|
||||
auto* state = new ompt_task_save_state{corr_id, flags};
|
||||
INTERNAL(new_task_data)->ptr = state;
|
||||
@@ -161,6 +162,7 @@ ompt_task_schedule_callback(ompt_data_t* prior_task_data,
|
||||
{
|
||||
auto* corr_id = ompt_impl<ROCPROFILER_OMPT_ID_task_schedule>::event_common(
|
||||
CLIENT(prior_task_data), prior_task_status, CLIENT(next_task_data));
|
||||
if(!corr_id) return; // During finalization
|
||||
context::pop_latest_correlation_id(corr_id);
|
||||
corr_id->sub_ref_count();
|
||||
|
||||
@@ -928,9 +930,13 @@ ompt_impl<OpIdx>::event_common(Args... args)
|
||||
buffered_contexts,
|
||||
external_corr_ids);
|
||||
|
||||
auto buffer_record = common::init_public_api_struct(buffer_ompt_record_t{});
|
||||
auto tracer_data = common::init_public_api_struct(callback_ompt_data_t{});
|
||||
auto* corr_id = tracing::correlation_service::construct(ref_count);
|
||||
auto buffer_record = common::init_public_api_struct(buffer_ompt_record_t{});
|
||||
auto tracer_data = common::init_public_api_struct(callback_ompt_data_t{});
|
||||
auto* corr_id = tracing::correlation_service::construct(ref_count);
|
||||
|
||||
// During finalization, correlation ID construction may return nullptr
|
||||
if(!corr_id) return nullptr;
|
||||
|
||||
uint64_t internal_corr_id = corr_id->internal;
|
||||
uint64_t ancestor_corr_id = corr_id->ancestor;
|
||||
|
||||
@@ -981,6 +987,7 @@ void
|
||||
ompt_impl<OpIdx>::event(Args&&... args)
|
||||
{
|
||||
auto corr_id = ompt_impl<OpIdx>::event_common(std::forward<Args>(args)...);
|
||||
if(!corr_id) return; // During finalization
|
||||
context::pop_latest_correlation_id(corr_id);
|
||||
corr_id->sub_ref_count();
|
||||
}
|
||||
|
||||
@@ -198,20 +198,18 @@ tool_tracing_buffered(rocprofiler_context_id_t context,
|
||||
<< ", drop_count=" << drop_count << ", start=" << record->start_timestamp
|
||||
<< ", stop=" << record->end_timestamp;
|
||||
|
||||
static int64_t last_corr_id = -1;
|
||||
auto corr_id = static_cast<int64_t>(record->correlation_id.internal);
|
||||
auto corr_id = static_cast<int64_t>(record->correlation_id.internal);
|
||||
|
||||
std::cout << info.str() << "\n" << std::flush;
|
||||
EXPECT_GE(context.handle, 0) << info.str();
|
||||
EXPECT_GT(record->thread_id, 0) << info.str();
|
||||
EXPECT_GT(record->kind, 0) << info.str();
|
||||
EXPECT_GT(corr_id, last_corr_id) << info.str();
|
||||
EXPECT_GT(corr_id, 0) << info.str();
|
||||
EXPECT_GT(record->start_timestamp, 0) << info.str();
|
||||
EXPECT_GT(record->end_timestamp, 0) << info.str();
|
||||
EXPECT_LE(record->start_timestamp, record->end_timestamp) << info.str();
|
||||
|
||||
cb_data->client_callback_count++;
|
||||
last_corr_id = corr_id;
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
@@ -0,0 +1,24 @@
|
||||
#
|
||||
# AddressSanitizer suppressions file for rocprofiler project.
|
||||
#
|
||||
# ASAN uses interceptor_via_lib and interceptor_via_fun for suppressing
|
||||
# issues in specific libraries or functions.
|
||||
#
|
||||
|
||||
# Suppress issues in HSA runtime library
|
||||
interceptor_via_lib:libhsa-runtime64.so
|
||||
|
||||
# Suppress issues in HIP runtime library
|
||||
interceptor_via_lib:libamdhip64.so
|
||||
|
||||
# Suppress issues in ROCm SMI library
|
||||
interceptor_via_lib:librocm_smi64.so
|
||||
|
||||
# Suppress issues in AQL profile library
|
||||
interceptor_via_lib:libhsa-amd-aqlprofile64.so
|
||||
|
||||
# Suppress issues in comgr library
|
||||
interceptor_via_lib:libamd_comgr.so
|
||||
|
||||
# Suppress issues in perfetto library (if linked dynamically)
|
||||
interceptor_via_lib:libperfetto.so
|
||||
|
||||
@@ -92,13 +92,11 @@ class FormatAll(argparse.Action):
|
||||
|
||||
class InstallDepsUbuntu(argparse.Action):
|
||||
def __call__(self, parser, namespace, values, option_string=None):
|
||||
os.system(
|
||||
"sudo apt-get update; \
|
||||
os.system("sudo apt-get update; \
|
||||
sudo apt-get install -y python3-pip software-properties-common wget curl clang-format-11; \
|
||||
python3 -m pip install -U cmake-format; \
|
||||
python -m pip install --upgrade pip; \
|
||||
python -m pip install black"
|
||||
)
|
||||
python -m pip install black")
|
||||
exit(0)
|
||||
|
||||
|
||||
|
||||
@@ -10,3 +10,18 @@ leak:hsa-amd-aqlprofile
|
||||
leak:__new_exitfn
|
||||
leak:omptarget
|
||||
leak:llvm
|
||||
leak:bash
|
||||
|
||||
# Suppress leaks from external tools and libraries
|
||||
leak:liblzma
|
||||
leak:liblsan
|
||||
leak:seq
|
||||
leak:__GI___strdup
|
||||
|
||||
# Suppress leaks during tool detach/shutdown - these allocations are
|
||||
# intentionally not freed as the process is exiting
|
||||
leak:rocprofiler::tool::metadata::get_string_entries
|
||||
leak:rocprofiler::tool::write_rocpd
|
||||
leak:rocprofiler::tool::generate_output
|
||||
leak:tool_detach
|
||||
leak:rocprofiler_register_detach
|
||||
|
||||
@@ -116,15 +116,39 @@ def generate_custom(args, cmake_args, ctest_args):
|
||||
MEMCHECK_SUPPRESSION_FILE = ""
|
||||
|
||||
if MEMCHECK_TYPE == "AddressSanitizer":
|
||||
MEMCHECK_SANITIZER_OPTIONS = "detect_leaks=0 use_sigaltstack=0"
|
||||
# print_suppressions=1 shows which suppressions matched during the run
|
||||
MEMCHECK_SANITIZER_OPTIONS = (
|
||||
"detect_leaks=0 use_sigaltstack=0 print_suppressions=1"
|
||||
)
|
||||
MEMCHECK_SUPPRESSION_FILE = (
|
||||
f"{SOURCE_DIR}/source/scripts/address-sanitizer-suppr.txt"
|
||||
)
|
||||
os.environ["ASAN_OPTIONS"] = " ".join(
|
||||
[
|
||||
"detect_leaks=0",
|
||||
"use_sigaltstack=0",
|
||||
"print_suppressions=1",
|
||||
f"suppressions={SOURCE_DIR}/source/scripts/address-sanitizer-suppr.txt",
|
||||
os.environ.get("ASAN_OPTIONS", ""),
|
||||
]
|
||||
)
|
||||
elif MEMCHECK_TYPE == "LeakSanitizer":
|
||||
# fast_unwind_on_malloc=1 avoids deadlock in libgcc unwinder during early init
|
||||
# print_suppressions=1 shows which suppressions matched during the run
|
||||
MEMCHECK_SANITIZER_OPTIONS = "fast_unwind_on_malloc=1 print_suppressions=1"
|
||||
MEMCHECK_SUPPRESSION_FILE = (
|
||||
f"{SOURCE_DIR}/source/scripts/leak-sanitizer-suppr.txt"
|
||||
)
|
||||
os.environ["LSAN_OPTIONS"] = " ".join(
|
||||
[
|
||||
f"suppressions={SOURCE_DIR}/source/scripts/leak-sanitizer-suppr.txt",
|
||||
"fast_unwind_on_malloc=1",
|
||||
"print_suppressions=1",
|
||||
os.environ.get("LSAN_OPTIONS", ""),
|
||||
]
|
||||
)
|
||||
elif MEMCHECK_TYPE == "ThreadSanitizer":
|
||||
# print_suppressions=1 shows which suppressions matched during the run
|
||||
external_symbolizer_path = ""
|
||||
for version in range(8, 20):
|
||||
_symbolizer = shutil.which(f"llvm-symbolizer-{version}")
|
||||
@@ -134,6 +158,7 @@ def generate_custom(args, cmake_args, ctest_args):
|
||||
[
|
||||
"history_size=5",
|
||||
"detect_deadlocks=0",
|
||||
"print_suppressions=1",
|
||||
f"suppressions={SOURCE_DIR}/source/scripts/thread-sanitizer-suppr.txt",
|
||||
external_symbolizer_path,
|
||||
os.environ.get("TSAN_OPTIONS", ""),
|
||||
@@ -151,6 +176,20 @@ def generate_custom(args, cmake_args, ctest_args):
|
||||
]
|
||||
)
|
||||
|
||||
# Print suppression file contents for debugging
|
||||
if MEMCHECK_TYPE:
|
||||
print(f"\n{'=' * 60}")
|
||||
print(f"Sanitizer: {MEMCHECK_TYPE}")
|
||||
print(f"{'=' * 60}")
|
||||
|
||||
# Print environment variables for sanitizers that use them
|
||||
for env_var in ["TSAN_OPTIONS", "UBSAN_OPTIONS", "ASAN_OPTIONS", "LSAN_OPTIONS"]:
|
||||
if env_var in os.environ:
|
||||
print(f"\n{env_var}:")
|
||||
print(f" {os.environ[env_var]}")
|
||||
|
||||
print(f"\n{'=' * 60}\n")
|
||||
|
||||
codecov_exclude = [
|
||||
"/usr/.*",
|
||||
"/opt/.*",
|
||||
|
||||
@@ -33,7 +33,7 @@ if [ -n "${EXTERNAL_SYMBOLIZER_PATH}" ]; then
|
||||
fi
|
||||
|
||||
: ${ASAN_OPTIONS="detect_leaks=0 use_sigaltstack=0 suppressions=${SUPPR_DIR}/address-sanitizer-suppr.txt"}
|
||||
: ${LSAN_OPTIONS="suppressions=${SUPPR_DIR}/leak-sanitizer-suppr.txt"}
|
||||
: ${LSAN_OPTIONS="fast_unwind_on_malloc=1 suppressions=${SUPPR_DIR}/leak-sanitizer-suppr.txt"}
|
||||
: ${TSAN_OPTIONS="history_size=5 detect_deadlocks=0 suppressions=${SUPPR_DIR}/thread-sanitizer-suppr.txt${EXTERNAL_SYMBOLIZER}"}
|
||||
: ${UBSAN_OPTIONS="print_stacktrace=1 suppressions=${SUPPR_DIR}/undef-behavior-sanitizer-suppr.txt${EXTERNAL_SYMBOLIZER}"}
|
||||
|
||||
|
||||
@@ -29,11 +29,24 @@ race:tzset_internal
|
||||
# double mutex lock (there isn't one)
|
||||
mutex:external/ptl/source/PTL/TaskGroup.hh
|
||||
|
||||
# data race in PTL TaskGroup between worker thread finishing and
|
||||
# main thread accessing shared_ptr in wait(). The shared_ptr refcount
|
||||
# may be accessed concurrently after wait() returns but before worker
|
||||
# thread fully completes its bookkeeping.
|
||||
race:rocprofiler::internal_threading::TaskGroup::wait
|
||||
race:PTL::PackagedTask
|
||||
|
||||
# lock order inversion that cannot happen
|
||||
mutex:source/lib/common/synchronized.hpp
|
||||
|
||||
# signal-unsafe function called from signal handler
|
||||
signal:rocprofv3_error_signal_handler
|
||||
|
||||
# data race within perfetto internals
|
||||
# data race within perfetto internals (includes TracingMuxer and protozero)
|
||||
race:perfetto::internal::
|
||||
race:perfetto::protos::
|
||||
race:protozero::
|
||||
|
||||
# False positive: C++11 guarantees thread-safe static local variable initialization.
|
||||
# TSAN incorrectly reports a race on the static initialization in create_write_functor.
|
||||
race:rocprofiler::hip::stream::create_write_functor
|
||||
|
||||
@@ -0,0 +1,66 @@
|
||||
#
|
||||
# UndefinedBehaviorSanitizer suppressions file for rocprofiler project.
|
||||
#
|
||||
# UBSAN suppressions use the format: type:pattern
|
||||
# where type can be: alignment, bool, bounds, enum, float-cast-overflow,
|
||||
# float-divide-by-zero, function, integer-divide-by-zero, null, object-size,
|
||||
# pointer-overflow, return, returns-nonnull-attribute, shift, signed-integer-overflow,
|
||||
# unreachable, unsigned-integer-overflow, vla-bound, vptr
|
||||
#
|
||||
|
||||
# Suppress undefined behavior in HSA runtime
|
||||
vptr:libhsa-runtime64.so
|
||||
function:libhsa-runtime64.so
|
||||
|
||||
# Suppress undefined behavior in HIP runtime
|
||||
vptr:libamdhip64.so
|
||||
function:libamdhip64.so
|
||||
|
||||
# Suppress undefined behavior in ROCm SMI
|
||||
vptr:librocm_smi64.so
|
||||
|
||||
# Suppress undefined behavior in comgr
|
||||
vptr:libamd_comgr.so
|
||||
|
||||
# Suppress alignment issues in external libraries
|
||||
alignment:libhsa-runtime64.so
|
||||
alignment:libamdhip64.so
|
||||
|
||||
# Suppress issues in perfetto (includes TracingMuxer and protozero)
|
||||
vptr:perfetto::
|
||||
function:perfetto::
|
||||
vptr:protozero::
|
||||
function:protozero::
|
||||
vptr:perfetto::internal::
|
||||
function:perfetto::internal::
|
||||
vptr:perfetto::protos::
|
||||
function:perfetto::protos::
|
||||
|
||||
# Suppress issues arising from OpenMP
|
||||
vptr:__kmp_resume_template
|
||||
function:__kmp_resume_template
|
||||
vptr:__kmp_suspend_initialize_thread
|
||||
function:__kmp_suspend_initialize_thread
|
||||
|
||||
# Suppress issues in google logging
|
||||
vptr:google::LogMessageTime::CalcGmtOffset
|
||||
function:google::LogMessageTime::CalcGmtOffset
|
||||
vptr:tzset_internal
|
||||
function:tzset_internal
|
||||
|
||||
# Suppress issues in PTL TaskGroup
|
||||
vptr:PTL::TaskGroup
|
||||
function:PTL::TaskGroup
|
||||
vptr:PTL::PackagedTask
|
||||
function:PTL::PackagedTask
|
||||
vptr:rocprofiler::internal_threading::TaskGroup
|
||||
function:rocprofiler::internal_threading::TaskGroup
|
||||
|
||||
# Suppress issues in synchronized.hpp
|
||||
vptr:source/lib/common/synchronized.hpp
|
||||
function:source/lib/common/synchronized.hpp
|
||||
|
||||
# Suppress issues in AQL profile library
|
||||
vptr:libhsa-amd-aqlprofile64.so
|
||||
function:libhsa-amd-aqlprofile64.so
|
||||
alignment:libhsa-amd-aqlprofile64.so
|
||||
|
||||
Fai riferimento in un nuovo problema
Block a user