Updates/fixes for CI, docs, tests, samples, and common library (#528)

- .github/workflows/continuous_integration.yml - apt-get update before apt-get install - remove libgtest-dev - actions-comment-pull-request: v2.4.3 -> v2.5.0 - .github/workflows/formatting.yml - create-pull-request: v5 -> v6 - cmake/rocprofiler_options.cmake - remove unused ROCPROFILER_DEBUG_TRACE and ROCPROFILER_LD_AQLPROFILE options - samples/counter_collection/callback_client.cpp - corr_id field renamed to correlation_id - samples/counter_collection/client.cpp - corr_id field renamed to correlation_id - include/rocprofiler-sdk/fwd.h - In rocprofiler_record_counter_t: rename corr_id field to correlation_id - doxygen fixes - lib/common/utility.* - remove get_accurate_clock_id_impl - timestamp_ns() defaults to CLOCK_BOOTTIME - lib/rocprofiler-sdk/counters/core.cpp - fix spelling mistake: extrenal -> external - corr_id field renamed to correlation_id - lib/rocprofiler-sdk-tool/tool.cpp - fix destruction of static tool::output_file before finalization - scripts/update-docs.sh - define PROJECT_NAME - tests/async-copy-tracing/validate.py - init_time and fini_time checks - hip_api_traces, marker_api_tracing - tests/common/serialization.hpp - fix save function for rocprofiler_record_counter_t following rename of corr_id to correlation_id - tests/kernel-tracing/validate.py - init_time and fini_time checks - relax test_total_runtime range - tests/rocprofv3/tracing/CMakeLists.txt - remove -M from rocprofv3-test-systrace-execute - exclude test_hsa_api_trace in rocprofv3-test-systrace-validate due to HIP API tracing - tests/rocprofv3/tracing/validate.py - update test_kernel_trace to accept mangled or demangled - tests/tools/json-tool.cpp - remove use of GLOG - include init_time and fini_time - write_json(...) function [ROCm/rocprofiler-sdk commit: 0d939edbba]
2024-02-22 00:16:43 -06:00
parent 2d9779ad96
commit f760a3ceaa
17 changed files with 292 additions and 233 deletions
@@ -8,7 +8,8 @@ import pytest
 def node_exists(name, data, min_len=1):
    assert name in data
    assert data[name] is not None
-    assert len(data[name]) >= min_len
+    if isinstance(data[name], (list, tuple, dict, set)):
+        assert len(data[name]) >= min_len


 def test_data_structure(input_data):
@@ -19,6 +20,12 @@ def test_data_structure(input_data):

    sdk_data = data["rocprofiler-sdk-json-tool"]

+    node_exists("metadata", sdk_data)
+    node_exists("pid", sdk_data["metadata"])
+    node_exists("main_tid", sdk_data["metadata"])
+    node_exists("init_time", sdk_data["metadata"])
+    node_exists("fini_time", sdk_data["metadata"])
+
    node_exists("agents", sdk_data)
    node_exists("call_stack", sdk_data)
    node_exists("callback_records", sdk_data)
@@ -28,11 +35,15 @@ def test_data_structure(input_data):
    node_exists("code_objects", sdk_data["callback_records"])
    node_exists("kernel_symbols", sdk_data["callback_records"])
    node_exists("hsa_api_traces", sdk_data["callback_records"])
+    node_exists("hip_api_traces", sdk_data["callback_records"], 0)
+    node_exists("marker_api_traces", sdk_data["callback_records"])

    node_exists("names", sdk_data["buffer_records"])
    node_exists("kernel_dispatches", sdk_data["buffer_records"])
    node_exists("memory_copies", sdk_data["buffer_records"], 4)
    node_exists("hsa_api_traces", sdk_data["buffer_records"])
+    node_exists("hip_api_traces", sdk_data["buffer_records"], 0)
+    node_exists("marker_api_traces", sdk_data["buffer_records"])


 def test_timestamps(input_data):
@@ -41,7 +52,7 @@ def test_timestamps(input_data):

    cb_start = {}
    cb_end = {}
-    for titr in ["hsa_api_traces", "marker_api_traces"]:
+    for titr in ["hsa_api_traces", "marker_api_traces", "hip_api_traces"]:
        for itr in sdk_data["callback_records"][titr]:
            cid = itr["record"]["correlation_id"]["internal"]
            phase = itr["record"]["phase"]
@@ -56,18 +67,20 @@ def test_timestamps(input_data):
        for itr in sdk_data["buffer_records"][titr]:
            assert itr["start_timestamp"] <= itr["end_timestamp"]

-    for itr in sdk_data["buffer_records"]["memory_copies"]:
-        assert itr["start_timestamp"] <= itr["end_timestamp"]
+    for titr in ["kernel_dispatches", "memory_copies"]:
+        for itr in sdk_data["buffer_records"][titr]:
+            assert itr["start_timestamp"] < itr["end_timestamp"]
+            assert itr["correlation_id"]["internal"] > 0
+            assert itr["correlation_id"]["external"] > 0
+            assert sdk_data["metadata"]["init_time"] < itr["start_timestamp"]
+            assert sdk_data["metadata"]["init_time"] < itr["end_timestamp"]
+            assert sdk_data["metadata"]["fini_time"] > itr["start_timestamp"]
+            assert sdk_data["metadata"]["fini_time"] > itr["end_timestamp"]

-    for itr in sdk_data["buffer_records"]["kernel_dispatches"]:
-        assert itr["start_timestamp"] < itr["end_timestamp"]
-        assert itr["correlation_id"]["internal"] > 0
-        assert itr["correlation_id"]["external"] > 0
-
-        api_start = cb_start[itr["correlation_id"]["internal"]]
-        api_end = cb_end[itr["correlation_id"]["internal"]]
-        assert api_start < itr["start_timestamp"]
-        assert api_end <= itr["end_timestamp"]
+            api_start = cb_start[itr["correlation_id"]["internal"]]
+            api_end = cb_end[itr["correlation_id"]["internal"]]
+            assert api_start < itr["start_timestamp"]
+            assert api_end <= itr["end_timestamp"]


 def test_internal_correlation_ids(input_data):
@@ -75,7 +88,7 @@ def test_internal_correlation_ids(input_data):
    sdk_data = data["rocprofiler-sdk-json-tool"]

    api_corr_ids = []
-    for titr in ["hsa_api_traces", "marker_api_traces"]:
+    for titr in ["hsa_api_traces", "marker_api_traces", "hip_api_traces"]:
        for itr in sdk_data["callback_records"][titr]:
            api_corr_ids.append(itr["record"]["correlation_id"]["internal"])

@@ -101,7 +114,7 @@ def test_external_correlation_ids(input_data):
    sdk_data = data["rocprofiler-sdk-json-tool"]

    extern_corr_ids = []
-    for titr in ["hsa_api_traces", "marker_api_traces"]:
+    for titr in ["hsa_api_traces", "marker_api_traces", "hip_api_traces"]:
        for itr in sdk_data["callback_records"][titr]:
            assert itr["record"]["correlation_id"]["external"] > 0
            assert (
@@ -110,7 +123,7 @@ def test_external_correlation_ids(input_data):
            extern_corr_ids.append(itr["record"]["correlation_id"]["external"])

    extern_corr_ids = list(set(sorted(extern_corr_ids)))
-    for titr in ["hsa_api_traces", "marker_api_traces"]:
+    for titr in ["hsa_api_traces", "marker_api_traces", "hip_api_traces"]:
        for itr in sdk_data["buffer_records"][titr]:
            assert itr["correlation_id"]["external"] > 0
            assert itr["thread_id"] == itr["correlation_id"]["external"]
@@ -205,9 +205,9 @@ save(ArchiveT& ar, rocprofiler_callback_tracing_record_t data)
 {
    SAVE_DATA_FIELD(context_id);
    SAVE_DATA_FIELD(thread_id);
-    SAVE_DATA_FIELD(correlation_id);
    SAVE_DATA_FIELD(kind);
    SAVE_DATA_FIELD(operation);
+    SAVE_DATA_FIELD(correlation_id);
    SAVE_DATA_FIELD(phase);
 }

@@ -217,8 +217,8 @@ save_buffer_tracing_api_record(ArchiveT& ar, Tp data)
 {
    SAVE_DATA_FIELD(size);
    SAVE_DATA_FIELD(kind);
-    SAVE_DATA_FIELD(correlation_id);
    SAVE_DATA_FIELD(operation);
+    SAVE_DATA_FIELD(correlation_id);
    SAVE_DATA_FIELD(start_timestamp);
    SAVE_DATA_FIELD(end_timestamp);
    SAVE_DATA_FIELD(thread_id);
@@ -237,7 +237,7 @@ save(ArchiveT& ar, rocprofiler_record_counter_t data)
 {
    SAVE_DATA_FIELD(id);
    SAVE_DATA_FIELD(counter_value);
-    SAVE_DATA_FIELD(corr_id);
+    SAVE_DATA_FIELD(correlation_id);
 }

 template <typename ArchiveT>
@@ -278,8 +278,8 @@ save(ArchiveT& ar, rocprofiler_buffer_tracing_memory_copy_record_t data)
 {
    SAVE_DATA_FIELD(size);
    SAVE_DATA_FIELD(kind);
-    SAVE_DATA_FIELD(correlation_id);
    SAVE_DATA_FIELD(operation);
+    SAVE_DATA_FIELD(correlation_id);
    SAVE_DATA_FIELD(start_timestamp);
    SAVE_DATA_FIELD(end_timestamp);
    SAVE_DATA_FIELD(dst_agent_id);
@@ -8,7 +8,8 @@ import pytest
 def node_exists(name, data, min_len=1):
    assert name in data
    assert data[name] is not None
-    assert len(data[name]) >= min_len
+    if isinstance(data[name], (list, tuple, dict, set)):
+        assert len(data[name]) >= min_len


 def test_data_structure(input_data):
@@ -19,6 +20,12 @@ def test_data_structure(input_data):

    sdk_data = data["rocprofiler-sdk-json-tool"]

+    node_exists("metadata", sdk_data)
+    node_exists("pid", sdk_data["metadata"])
+    node_exists("main_tid", sdk_data["metadata"])
+    node_exists("init_time", sdk_data["metadata"])
+    node_exists("fini_time", sdk_data["metadata"])
+
    node_exists("agents", sdk_data)
    node_exists("call_stack", sdk_data)
    node_exists("callback_records", sdk_data)
@@ -60,18 +67,20 @@ def test_timestamps(input_data):
        for itr in sdk_data["buffer_records"][titr]:
            assert itr["start_timestamp"] <= itr["end_timestamp"]

-    for itr in sdk_data["buffer_records"]["memory_copies"]:
-        assert itr["start_timestamp"] <= itr["end_timestamp"]
+    for titr in ["kernel_dispatches", "memory_copies"]:
+        for itr in sdk_data["buffer_records"][titr]:
+            assert itr["start_timestamp"] < itr["end_timestamp"]
+            assert itr["correlation_id"]["internal"] > 0
+            assert itr["correlation_id"]["external"] > 0
+            assert sdk_data["metadata"]["init_time"] < itr["start_timestamp"]
+            assert sdk_data["metadata"]["init_time"] < itr["end_timestamp"]
+            assert sdk_data["metadata"]["fini_time"] > itr["start_timestamp"]
+            assert sdk_data["metadata"]["fini_time"] > itr["end_timestamp"]

-    for itr in sdk_data["buffer_records"]["kernel_dispatches"]:
-        assert itr["start_timestamp"] < itr["end_timestamp"]
-        assert itr["correlation_id"]["internal"] > 0
-        assert itr["correlation_id"]["external"] > 0
-
-        api_start = cb_start[itr["correlation_id"]["internal"]]
-        api_end = cb_end[itr["correlation_id"]["internal"]]
-        assert api_start < itr["start_timestamp"]
-        assert api_end <= itr["end_timestamp"]
+            api_start = cb_start[itr["correlation_id"]["internal"]]
+            api_end = cb_end[itr["correlation_id"]["internal"]]
+            assert api_start < itr["start_timestamp"]
+            assert api_end <= itr["end_timestamp"]


 def test_total_runtime(input_data):
@@ -84,8 +93,8 @@ def test_total_runtime(input_data):

    expected_runtime = 1.0e3  # one second in milliseconds

-    assert (sum(runtime_data) * 1.0e-6) >= (0.9 * expected_runtime)
-    assert (sum(runtime_data) * 1.0e-6) <= (1.1 * expected_runtime)
+    assert (sum(runtime_data) * 1.0e-6) >= (0.8 * expected_runtime)
+    assert (sum(runtime_data) * 1.0e-6) <= (1.2 * expected_runtime)


 def test_internal_correlation_ids(input_data):
@@ -84,7 +84,7 @@ set_tests_properties(
 add_test(
    NAME rocprofv3-test-systrace-execute
    COMMAND
-        $<TARGET_FILE:rocprofiler-sdk::rocprofv3> -M --sys-trace -d
+        $<TARGET_FILE:rocprofiler-sdk::rocprofv3> --sys-trace -d
        ${CMAKE_CURRENT_BINARY_DIR}/%argt%-systrace -o out
        $<TARGET_FILE:simple-transpose>)

@@ -102,7 +102,8 @@ set_tests_properties(
 add_test(
    NAME rocprofv3-test-systrace-validate
    COMMAND
-        ${Python3_EXECUTABLE} ${CMAKE_CURRENT_BINARY_DIR}/validate.py --hsa-input
+        ${Python3_EXECUTABLE} ${CMAKE_CURRENT_BINARY_DIR}/validate.py -k
+        "not test_hsa_api_trace" --hsa-input
        ${CMAKE_CURRENT_BINARY_DIR}/simple-transpose-systrace/out_hsa_api_trace.csv
        --kernel-input
        ${CMAKE_CURRENT_BINARY_DIR}/simple-transpose-systrace/out_kernel_trace.csv
@@ -2,7 +2,6 @@

 import sys
 import pytest
-import re


 def test_hsa_api_trace(hsa_input_data):
@@ -37,13 +36,18 @@ def test_hsa_api_trace(hsa_input_data):


 def test_kernel_trace(kernel_input_data):
-    mangled_kernel_name = "_Z15matrixTransposePfS_i.kd"
+    valid_kernel_names = (
+        "_Z15matrixTransposePfS_i.kd",
+        "matrixTranspose(float*, float*, int)",
+    )
+
+    assert len(kernel_input_data) == 1
    for row in kernel_input_data:
        assert row["Kind"] == "KERNEL_DISPATCH"
        assert int(row["Agent_Id"]) > 0
        assert int(row["Queue_Id"]) > 0
        assert int(row["Kernel_Id"]) > 0
-        assert row["Kernel_Name"] == mangled_kernel_name
+        assert row["Kernel_Name"] in valid_kernel_names
        assert int(row["Correlation_Id"]) > 0
        assert int(row["Workgroup_Size_X"]) == 4
        assert int(row["Workgroup_Size_Y"]) == 4
@@ -43,8 +43,6 @@
 #include <rocprofiler-sdk/registration.h>
 #include <rocprofiler-sdk/rocprofiler.h>

-#include <glog/logging.h>
-
 #include <unistd.h>
 #include <atomic>
 #include <cassert>
@@ -432,7 +430,7 @@ dispatch_callback(rocprofiler_queue_id_t, /*queue_id*/
    // Counters we want to collect (here its SQ_WAVES_sum)
    auto* counters_env = getenv("ROCPROF_COUNTERS");
    if(std::string(counters_env) != "SQ_WAVES_sum")
-        LOG(FATAL) << "Counter not supported in the test tool";
+        throw std::runtime_error{"Counter not supported in the test tool"};

    std::set<std::string> counters_to_collect = {"SQ_WAVES_sum"};
    // GPU Counter IDs
@@ -728,9 +726,16 @@ auto buffers = std::array<rocprofiler_buffer_id_t*, 6>{&hsa_api_buffered_buffer,

 auto agents = std::vector<rocprofiler_agent_t>{};

+rocprofiler_timestamp_t init_time = 0;
+rocprofiler_timestamp_t fini_time = 0;
+rocprofiler_thread_id_t main_tid  = 0;
+
 int
 tool_init(rocprofiler_client_finalize_t fini_func, void* tool_data)
 {
+    rocprofiler_get_timestamp(&init_time);
+    rocprofiler_get_thread_id(&main_tid);
+
    assert(tool_data != nullptr);

    rocprofiler_available_agents_cb_t iterate_cb =
@@ -997,6 +1002,9 @@ tool_init(rocprofiler_client_finalize_t fini_func, void* tool_data)
    return 0;
 }

+void
+write_json(call_stack_t* _call_stack);
+
 void
 tool_fini(void* tool_data)
 {
@@ -1006,6 +1014,8 @@ tool_fini(void* tool_data)
    stop();
    flush();

+    rocprofiler_get_timestamp(&fini_time);
+
    std::cerr << "[" << getpid() << "][" << __FUNCTION__
              << "] Finalizing... agents=" << agents.size()
              << ", code_object_callback_records=" << code_object_records.size()
@@ -1027,6 +1037,17 @@ tool_fini(void* tool_data)
        _call_stack->emplace_back(source_location{__FUNCTION__, __FILE__, __LINE__, ""});
    }

+    write_json(_call_stack);
+
+    std::cerr << "[" << getpid() << "][" << __FUNCTION__ << "] Finalization complete.\n"
+              << std::flush;
+
+    delete _call_stack;
+}
+
+void
+write_json(call_stack_t* _call_stack)
+{
    auto ofname = std::string{"rocprofiler-tool-results.json"};
    if(auto* eofname = getenv("ROCPROFILER_TOOL_OUTPUT_FILE")) ofname = eofname;

@@ -1068,6 +1089,14 @@ tool_fini(void* tool_data)
        json_ar.setNextName("rocprofiler-sdk-json-tool");
        json_ar.startNode();

+        json_ar.setNextName("metadata");
+        json_ar.startNode();
+        json_ar(cereal::make_nvp("pid", getpid()));
+        json_ar(cereal::make_nvp("main_tid", main_tid));
+        json_ar(cereal::make_nvp("init_time", init_time));
+        json_ar(cereal::make_nvp("fini_time", fini_time));
+        json_ar.finishNode();
+
        json_ar(cereal::make_nvp("agents", agents));
        if(_call_stack) json_ar(cereal::make_nvp("call_stack", *_call_stack));

@@ -1114,11 +1143,6 @@ tool_fini(void* tool_data)
    *ofs << std::flush;

    if(cleanup) cleanup(ofs);
-
-    std::cerr << "[" << getpid() << "][" << __FUNCTION__ << "] Finalization complete.\n"
-              << std::flush;
-
-    delete _call_stack;
 }

 void