Shared Library Constructor (rocprofv3 deadlock fix) (#599)

* Moved tests/apps to tests/bin * Renamed cmake project in tests/bin * Update samples - Use ROCPROFILER_DEFAULT_FAIL_REGEX - tweaks to stdout messages * Update tests - Use ROCPROFILER_DEFAULT_FAIL_REGEX * Add tests/lib - libraries with HIP code * Update PTL submodule - remove atexit delete of thread_id_map * Update cmake/rocprofiler_options.cmake - Set ROCPROFILER_DEFAULT_FAIL_REGEX * Update common lib: env + logging - improved customization of logging settings - default to disabling logging to files - install failure handler for rocprofv3 - set_env support in environment.* * Add lib/rocprofiler-sdk/shared_library.cpp - shared library constructor * Update lib/rocprofiler-sdk-tool/tool.cpp - destructor thread safety - convert callback_name_info and buffered_name_info to pointers - install failure handler for logging * Add tests/bin/hip-in-libraries - hip-in-libraries is an exe which uses two shared libraries where each shared library contains HIP kernels - used for testing deadlocking within __hipRegisterFatBinary * Update bin/rocprofv3 - reorganized the env variables - use exec to launch command - set ROCPROFILER_LIBRARY_CTOR=1 * Add tests/rocprofv3/tracing-hip-in-libraries - uses hip-in-libraries exe for exe which uses shared libraries to launch HIP kernels * Update bin/rocprofv3 - fix counter collection (no exec) * Update lib/rocprofiler-sdk-tool/tool.cpp - replace "Kernel-Name" with "Kernel_Name" * Update lib/rocprofiler-sdk/registration.cpp Use RTLD_LOCAL instead of RTLD_GLOBAL for env libraries * Update tests/rocprofv3 - replace "Kernel-Name" with "Kernel_Name" * Update tests - vector-ops (bin) stream syncs + runs with 4 queues per device - improve counter-collection/input1 validation - rocprofv3/tracing-hip-in-libraries does not do sys-trace - improved validation script for tracing-hip-in-libraries - updated dispatch_callback in json-tool.cpp following reworking of prototypes for counter collection * Update samples/counter_collection - updated dispatch_callback(s) and record_callback(s) following reworking of prototypes * Update bin/rocprofv3 - reorganized help menu - added options for sub-HSA tables - added --hip-runtime-trace - changed --hip-trace to include --hip-compiler-trace * Update lib/rocprofiler-sdk-tool - improved kernel filtering - removed arch_vgpr, accum_vgpr, sgpr code (in rocprofiler-sdk) - fixed issue with counter-collection w/o tracing - added support for fine grained HSA API tracing - removed directly linking to HSA-runtime * Update lib/rocprofiler-sdk/agent.cpp - rocp_agents != hsa_agents is non-fatal when ROCPROFILER_BUILD_CI=OFF (CMake option) * GPR (vector and scalar) info in kernel symbol data - rocprofiler_callback_tracing_code_object_kernel_symbol_register_data_t contains general purpose register info * Header include order fix - Include repo headers first - Third party library headers next - standard library headers last * Update dispatch profiling public API - introduce rocprofiler_profile_counting_dispatch_data_t - change signature of rocprofiler_profile_counting_dispatch_callback_t and rocprofiler_profile_counting_record_callback_t - provide rocprofiler_user_data_t pointer in dispatch callback - provide rocprofiler_user_data_t value (from dispatch cb) in record callback * Update tests/bin/CMakeLists.txt - fix add_subdirectory(hip-in-libraries) order * Update VERSION - bump to 0.2.0 in prep for AFAR [ROCm/rocprofiler-sdk commit: 7b6d3c70bd]
2024-03-07 22:21:26 -06:00
@@ -52,7 +52,7 @@ set_tests_properties(
        ENVIRONMENT
        "${ROCPROFILER_MEMCHECK_PRELOAD_ENV};HSA_TOOLS_LIB=$<TARGET_FILE:rocprofiler::rocprofiler-shared-library>"
        FAIL_REGULAR_EXPRESSION
-        "threw an exception")
+        "${ROCPROFILER_DEFAULT_FAIL_REGEX}")

 add_library(counter-collection-callback-client SHARED)
 target_sources(counter-collection-callback-client PRIVATE callback_client.cpp client.hpp)
@@ -80,7 +80,7 @@ set_tests_properties(
        ENVIRONMENT
        "${ROCPROFILER_MEMCHECK_PRELOAD_ENV};HSA_TOOLS_LIB=$<TARGET_FILE:rocprofiler::rocprofiler-shared-library>"
        FAIL_REGULAR_EXPRESSION
-        "threw an exception")
+        "${ROCPROFILER_DEFAULT_FAIL_REGEX}")

 add_library(counter-collection-functional-counter-client SHARED)
 target_sources(counter-collection-functional-counter-client
@@ -109,4 +109,4 @@ set_tests_properties(
        ENVIRONMENT
        "${ROCPROFILER_MEMCHECK_PRELOAD_ENV};HSA_TOOLS_LIB=$<TARGET_FILE:rocprofiler::rocprofiler-shared-library>"
        FAIL_REGULAR_EXPRESSION
-        "threw an exception")
+        "${ROCPROFILER_DEFAULT_FAIL_REGEX}")
@@ -67,15 +67,14 @@ get_client_ctx()
 }

 void
-record_callback(rocprofiler_queue_id_t,
-                rocprofiler_agent_id_t,
-                rocprofiler_correlation_id_t,
-                uint64_t,
-                void*                         callback_data_args,
-                size_t                        record_count,
-                rocprofiler_record_counter_t* record_data)
+record_callback(rocprofiler_profile_counting_dispatch_data_t dispatch_data,
+                rocprofiler_record_counter_t*                record_data,
+                size_t                                       record_count,
+                rocprofiler_user_data_t                      user_data,
+                void*                                        callback_data_args)
 {
    std::stringstream ss;
+    ss << "Kernel_id " << dispatch_data.kernel_id << ": ";
    for(size_t i = 0; i < record_count; ++i)
    {
        ss << "(Id: " << record_data[i].id << " Value [D]: " << record_data[i].counter_value
@@ -84,6 +83,8 @@ record_callback(rocprofiler_queue_id_t,
    auto* output_stream = static_cast<std::ostream*>(callback_data_args);
    if(!output_stream) throw std::runtime_error{"nullptr to output stream"};
    *output_stream << "[" << __FUNCTION__ << "] " << ss.str() << "\n";
+
+    (void) user_data;
 }

 /**
@@ -93,13 +94,10 @@ record_callback(rocprofiler_queue_id_t,
 * to collect the counter SQ_WAVES for all kernel dispatch packets.
 */
 void
-dispatch_callback(rocprofiler_queue_id_t /*queue_id*/,
-                  const rocprofiler_agent_t* agent,
-                  rocprofiler_correlation_id_t /*correlation_id*/,
-                  const hsa_kernel_dispatch_packet_t* /*dispatch_packet*/,
-                  uint64_t /*kernel_id*/,
-                  void* /*callback_data_args*/,
-                  rocprofiler_profile_config_id_t* config)
+dispatch_callback(rocprofiler_profile_counting_dispatch_data_t dispatch_data,
+                  rocprofiler_profile_config_id_t*             config,
+                  rocprofiler_user_data_t* /*user_data*/,
+                  void* /*callback_data_args*/)
 {
    /**
     * This simple example uses the same profile counter set for all agents.
@@ -112,7 +110,7 @@ dispatch_callback(rocprofiler_queue_id_t /*queue_id*/,
    static std::unordered_map<uint64_t, rocprofiler_profile_config_id_t> profile_cache = {};

    auto search_cache = [&]() {
-        if(auto pos = profile_cache.find(agent->id.handle); pos != profile_cache.end())
+        if(auto pos = profile_cache.find(dispatch_data.agent_id.handle); pos != profile_cache.end())
        {
            *config = pos->second;
            return true;
@@ -135,7 +133,7 @@ dispatch_callback(rocprofiler_queue_id_t /*queue_id*/,

    // Iterate through the agents and get the counters available on that agent
    ROCPROFILER_CALL(rocprofiler_iterate_agent_supported_counters(
-                         agent->id,
+                         dispatch_data.agent_id,
                         [](rocprofiler_agent_id_t,
                            rocprofiler_counter_id_t* counters,
                            size_t                    num_counters,
@@ -169,11 +167,12 @@ dispatch_callback(rocprofiler_queue_id_t /*queue_id*/,

    // Create a colleciton profile for the counters
    rocprofiler_profile_config_id_t profile;
-    ROCPROFILER_CALL(rocprofiler_create_profile_config(
-                         agent->id, collect_counters.data(), collect_counters.size(), &profile),
-                     "Could not construct profile cfg");
+    ROCPROFILER_CALL(
+        rocprofiler_create_profile_config(
+            dispatch_data.agent_id, collect_counters.data(), collect_counters.size(), &profile),
+        "Could not construct profile cfg");

-    profile_cache.emplace(agent->id.handle, profile);
+    profile_cache.emplace(dispatch_data.agent_id.handle, profile);
    // Return the profile to collect those counters for this dispatch
    *config = profile;
 }
@@ -32,6 +32,7 @@
 #include <unordered_map>
 #include <vector>

+#include <rocprofiler-sdk/fwd.h>
 #include <rocprofiler-sdk/registration.h>
 #include <rocprofiler-sdk/rocprofiler.h>

@@ -116,13 +117,10 @@ buffered_callback(rocprofiler_context_id_t,
 * to collect the counter SQ_WAVES for all kernel dispatch packets.
 */
 void
-dispatch_callback(rocprofiler_queue_id_t /*queue_id*/,
-                  const rocprofiler_agent_t* agent,
-                  rocprofiler_correlation_id_t /*correlation_id*/,
-                  const hsa_kernel_dispatch_packet_t* /*dispatch_packet*/,
-                  uint64_t /*kernel_id*/,
-                  void* /*callback_data_args*/,
-                  rocprofiler_profile_config_id_t* config)
+dispatch_callback(rocprofiler_profile_counting_dispatch_data_t dispatch_data,
+                  rocprofiler_profile_config_id_t*             config,
+                  rocprofiler_user_data_t* /*user_data*/,
+                  void* /*callback_data_args*/)
 {
    /**
     * This simple example uses the same profile counter set for all agents.
@@ -135,7 +133,7 @@ dispatch_callback(rocprofiler_queue_id_t /*queue_id*/,
    static std::unordered_map<uint64_t, rocprofiler_profile_config_id_t> profile_cache = {};

    auto search_cache = [&]() {
-        if(auto pos = profile_cache.find(agent->id.handle); pos != profile_cache.end())
+        if(auto pos = profile_cache.find(dispatch_data.agent_id.handle); pos != profile_cache.end())
        {
            *config = pos->second;
            return true;
@@ -158,7 +156,7 @@ dispatch_callback(rocprofiler_queue_id_t /*queue_id*/,

    // Iterate through the agents and get the counters available on that agent
    ROCPROFILER_CALL(rocprofiler_iterate_agent_supported_counters(
-                         agent->id,
+                         dispatch_data.agent_id,
                         [](rocprofiler_agent_id_t,
                            rocprofiler_counter_id_t* counters,
                            size_t                    num_counters,
@@ -192,11 +190,12 @@ dispatch_callback(rocprofiler_queue_id_t /*queue_id*/,

    // Create a colleciton profile for the counters
    rocprofiler_profile_config_id_t profile;
-    ROCPROFILER_CALL(rocprofiler_create_profile_config(
-                         agent->id, collect_counters.data(), collect_counters.size(), &profile),
-                     "Could not construct profile cfg");
+    ROCPROFILER_CALL(
+        rocprofiler_create_profile_config(
+            dispatch_data.agent_id, collect_counters.data(), collect_counters.size(), &profile),
+        "Could not construct profile cfg");

-    profile_cache.emplace(agent->id.handle, profile);
+    profile_cache.emplace(dispatch_data.agent_id.handle, profile);
    // Return the profile to collect those counters for this dispatch
    *config = profile;
 }
@@ -189,15 +189,46 @@ buffered_callback(rocprofiler_context_id_t,
    }
 }

-void
-dispatch_callback(rocprofiler_queue_id_t /*queue_id*/,
-                  const rocprofiler_agent_t* agent,
-                  rocprofiler_correlation_id_t /*correlation_id*/,
-                  const hsa_kernel_dispatch_packet_t* /*dispatch_packet*/,
-                  uint64_t /*kernel_id*/,
-                  void* /*callback_data_args*/,
-                  rocprofiler_profile_config_id_t* config)
+using agent_map_t = std::map<uint64_t, const rocprofiler_agent_v0_t*>;
+
+agent_map_t
+get_agent_info()
 {
+    auto iterate_cb = [](rocprofiler_agent_version_t agents_ver,
+                         const void**                agents_arr,
+                         size_t                      num_agents,
+                         void*                       user_data) {
+        if(agents_ver != ROCPROFILER_AGENT_INFO_VERSION_0)
+            throw std::runtime_error{"unexpected rocprofiler agent version"};
+
+        auto* agents_v = static_cast<agent_map_t*>(user_data);
+        for(size_t i = 0; i < num_agents; ++i)
+        {
+            const auto* itr = static_cast<const rocprofiler_agent_v0_t*>(agents_arr[i]);
+            agents_v->emplace(itr->id.handle, itr);
+        }
+        return ROCPROFILER_STATUS_SUCCESS;
+    };
+
+    auto _agents = agent_map_t{};
+    ROCPROFILER_CALL(
+        rocprofiler_query_available_agents(ROCPROFILER_AGENT_INFO_VERSION_0,
+                                           iterate_cb,
+                                           sizeof(rocprofiler_agent_t),
+                                           const_cast<void*>(static_cast<const void*>(&_agents))),
+        "query available agents");
+
+    return _agents;
+}
+
+void
+dispatch_callback(rocprofiler_profile_counting_dispatch_data_t dispatch_data,
+                  rocprofiler_profile_config_id_t*             config,
+                  rocprofiler_user_data_t* /*user_data*/,
+                  void* /*callback_data_args*/)
+{
+    static auto agents = get_agent_info();
+
    auto& cap   = *get_capture();
    auto  wlock = std::unique_lock{cap.m_mutex};

@@ -211,7 +242,7 @@ dispatch_callback(rocprofiler_queue_id_t /*queue_id*/,
    {
        std::vector<rocprofiler_counter_id_t> counters_needed;
        ROCPROFILER_CALL(rocprofiler_iterate_agent_supported_counters(
-                             agent->id,
+                             dispatch_data.agent_id,
                             [](rocprofiler_agent_id_t,
                                rocprofiler_counter_id_t* counters,
                                size_t                    num_counters,
@@ -237,9 +268,9 @@ dispatch_callback(rocprofiler_queue_id_t /*queue_id*/,
                             "Could not query counter_id");
            cap.expected_counter_names.emplace(found_counter.handle, std::string(version.name));
            size_t expected = 0;
-            ROCPROFILER_CALL(
-                rocprofiler_query_counter_instance_count(agent->id, found_counter, &expected),
-                "COULD NOT QUERY INSTANCES");
+            ROCPROFILER_CALL(rocprofiler_query_counter_instance_count(
+                                 dispatch_data.agent_id, found_counter, &expected),
+                             "COULD NOT QUERY INSTANCES");
            cap.remaining.push_back(found_counter);
            cap.expected.emplace(found_counter.handle, expected);

@@ -266,7 +297,8 @@ dispatch_callback(rocprofiler_queue_id_t /*queue_id*/,
        }
        if(cap.expected.empty())
        {
-            std::clog << "No counters found for agent - " << agent->name;
+            std::clog << "No counters found for agent " << dispatch_data.agent_id.handle << " ("
+                      << agents.at(dispatch_data.agent_id.handle)->name << ")";
        }
    }
    if(cap.remaining.empty()) return;
@@ -274,9 +306,9 @@ dispatch_callback(rocprofiler_queue_id_t /*queue_id*/,
    rocprofiler_profile_config_id_t profile;

    // Select the next counter to collect.
-    ROCPROFILER_CALL(
-        rocprofiler_create_profile_config(agent->id, &(cap.remaining.back()), 1, &profile),
-        "Could not construct profile cfg");
+    ROCPROFILER_CALL(rocprofiler_create_profile_config(
+                         dispatch_data.agent_id, &(cap.remaining.back()), 1, &profile),
+                     "Could not construct profile cfg");

    cap.remaining.pop_back();
    *config = profile;