[RSERP-1802] Add trace decoder to API (#398)

* Add trace decoder to API.

* Cleanup and activity

* Rename

* Minor fix

* Replace tt/TT with thread_trace/THREAD_TRACE

- public API types are not abbreviated

* Fix aliases

* Build system updates

- activate clang-tidy for all subfolders in lib
- fix addition of sources for att-tool

* Fix clang-tidy issues with lib/att-tool/counters.{hpp,cpp}

* Delete counters.cpp

* Formatting

---------

Co-authored-by: Giovanni Baraldi <gbaraldi@amd.com>
Co-authored-by: Jonathan R. Madsen <jonathanrmadsen@gmail.com>
Αυτή η υποβολή περιλαμβάνεται σε:
authored and GitHub committed 2025-05-17 12:08:33 -07:00
1 γονέας 3eb921a34c
υποβολή 65786f619d
56 files changed
+1175 -1419

No files matched your search

@@ -38,9 +38,7 @@ add_subdirectory(pc-sampling)
add_subdirectory(collection-period)
add_subdirectory(rocdecode-trace)
add_subdirectory(rocjpeg-trace)
if(TARGET att_decoder_testing1)
add_subdirectory(advanced-thread-trace)
endif()
add_subdirectory(advanced-thread-trace)
add_subdirectory(hip-stream-display)
add_subdirectory(agent-index)
add_subdirectory(negate-aggregate-tracing-options)
@@ -48,7 +48,7 @@ find_library(
HINTS ${LIB_PATH_LOC}
PATHS ${ROCM_PATH}
PATH_SUFFIXES lib
NAMES att_decoder_trace)
NAMES rocprof-trace-decoder)
if(attdecoder)
set(IS_DISABLED OFF)
@@ -62,7 +62,7 @@ add_test(
1 --att-target-cu 1 --att-shader-engine-mask 0x11 --kernel-include-regex copyD
--att-buffer-size 0x6000000 --att-simd-select 0x3 --att-serialize-all 1 -d
${CMAKE_CURRENT_BINARY_DIR}/%argt%-trace/cmd_input -o out --output-format json
${PRELOAD_ARGS} --att-library-path ${LIB_PATH_LOC} --
${PRELOAD_ARGS} --att-library-path ${LIB_PATH_LOC} --att-activity 8 --
$<TARGET_FILE:hsa_code_object_testapp>)
set_tests_properties(
@@ -164,7 +164,7 @@ add_test(
set_tests_properties(
rocprofv3-test-att-yaml-input-will-fail
PROPERTIES TIMEOUT 45 LABELS "integration-tests" WILL_FAIL ON DISABLED ${IS_DISABLED})
PROPERTIES TIMEOUT 45 LABELS "integration-tests" WILL_FAIL ON DISABLED True)
add_test(NAME rocprofv3-test-att-library-path-cmd-line-will-fail
COMMAND $<TARGET_FILE:rocprofiler-sdk::rocprofv3> --att --att-library-path .
@@ -172,7 +172,7 @@ add_test(NAME rocprofv3-test-att-library-path-cmd-line-will-fail
set_tests_properties(
rocprofv3-test-att-library-path-cmd-line-will-fail
PROPERTIES TIMEOUT 45 LABELS "integration-tests" WILL_FAIL ON DISABLED ${IS_DISABLED})
PROPERTIES TIMEOUT 45 LABELS "integration-tests" WILL_FAIL ON DISABLED True)
add_test(NAME rocprofv3-test-att-library-path-env-var-will-fail
COMMAND $<TARGET_FILE:rocprofiler-sdk::rocprofv3> --att --log-level env --echo
@@ -189,7 +189,7 @@ set_tests_properties(
WILL_FAIL
ON
DISABLED
${IS_DISABLED})
True)
#
# Uses ATT and Counter Collection at the same time
@@ -205,3 +205,45 @@ set_tests_properties(
rocprofv3-test-hsa-multiqueue-att-plus-pmc-execute
PROPERTIES TIMEOUT 45 LABELS "integration-tests" DISABLED ${IS_DISABLED} ENVIRONMENT
${LIB_PATH_ENV})
# Check for conflict PMC + activity
add_test(
NAME rocprofv3-test-hsa-multiqueue-att-activity-pmc-will-fail
COMMAND
$<TARGET_FILE:rocprofiler-sdk::rocprofv3> --log-level env --pmc SQ_WAVES
--advanced-thread-trace -d ${CMAKE_CURRENT_BINARY_DIR}/%argt%-trace/cmd_input -o
out --att-activity 8 ${PRELOAD_ARGS} -- $<TARGET_FILE:vector-ops>)
set_tests_properties(
rocprofv3-test-hsa-multiqueue-att-activity-pmc-will-fail
PROPERTIES TIMEOUT
45
LABELS
"integration-tests"
DISABLED
${IS_DISABLED}
ENVIRONMENT
${LIB_PATH_ENV}
WILL_FAIL
ON)
# Check for conflict Perfcounters + activity
add_test(
NAME rocprofv3-test-hsa-multiqueue-att-activity-perf-will-fail
COMMAND
$<TARGET_FILE:rocprofiler-sdk::rocprofv3> --log-level env --att
--att-perfcounter-ctrl 8 -d ${CMAKE_CURRENT_BINARY_DIR}/%argt%-trace/cmd_input -o
out --att-activity 8 ${PRELOAD_ARGS} -- $<TARGET_FILE:vector-ops>)
set_tests_properties(
rocprofv3-test-hsa-multiqueue-att-activity-perf-will-fail
PROPERTIES TIMEOUT
45
LABELS
"integration-tests"
DISABLED
${IS_DISABLED}
ENVIRONMENT
${LIB_PATH_ENV}
WILL_FAIL
ON)
@@ -65,8 +65,8 @@ dispatch_tracing_callback(rocprofiler_callback_tracing_record_t record,
if(var) return atoi(var);
return def;
};
static int begin_dispatch = get_int_var("ROCPROFILER_ATT_BEGIN", 1);
static int end_dispatch = get_int_var("ROCPROFILER_ATT_END", 4);
static int begin_dispatch = get_int_var("ROCPROFILER_THREAD_TRACE_BEGIN", 1);
static int end_dispatch = get_int_var("ROCPROFILER_THREAD_TRACE_END", 4);
static std::atomic<bool> isprofiling{false};
static std::mutex mut;
@@ -115,12 +115,12 @@ query_available_agents(rocprofiler_agent_version_t /* version */,
const auto* agent = static_cast<const rocprofiler_agent_v0_t*>(agents[idx]);
if(agent->type != ROCPROFILER_AGENT_TYPE_GPU) continue;
std::vector<rocprofiler_att_parameter_t> parameters;
parameters.push_back({ROCPROFILER_ATT_PARAMETER_TARGET_CU, 1});
parameters.push_back({ROCPROFILER_ATT_PARAMETER_SIMD_SELECT, 0xF});
parameters.push_back({ROCPROFILER_ATT_PARAMETER_BUFFER_SIZE, 0x6000000});
parameters.push_back({ROCPROFILER_ATT_PARAMETER_SHADER_ENGINE_MASK, 0x11});
parameters.push_back({ROCPROFILER_ATT_PARAMETER_SERIALIZE_ALL, 0});
std::vector<rocprofiler_thread_trace_parameter_t> parameters;
parameters.push_back({ROCPROFILER_THREAD_TRACE_PARAMETER_TARGET_CU, 1});
parameters.push_back({ROCPROFILER_THREAD_TRACE_PARAMETER_SIMD_SELECT, 0xF});
parameters.push_back({ROCPROFILER_THREAD_TRACE_PARAMETER_BUFFER_SIZE, 0x6000000});
parameters.push_back({ROCPROFILER_THREAD_TRACE_PARAMETER_SHADER_ENGINE_MASK, 0x11});
parameters.push_back({ROCPROFILER_THREAD_TRACE_PARAMETER_SERIALIZE_ALL, 0});
ROCPROFILER_CALL(
rocprofiler_configure_device_thread_trace_service(agent_ctx,
@@ -36,7 +36,7 @@ namespace Multi
{
rocprofiler_client_id_t* client_id = nullptr;
rocprofiler_att_control_flags_t
rocprofiler_thread_trace_control_flags_t
dispatch_callback(rocprofiler_agent_id_t /* agent */,
rocprofiler_queue_id_t /* queue_id */,
rocprofiler_async_correlation_id_t /* correlation_id */,
@@ -46,12 +46,12 @@ dispatch_callback(rocprofiler_agent_id_t /* agent */,
rocprofiler_user_data_t* dispatch_userdata)
{
static std::atomic<size_t> count{0};
if(count.fetch_add(1) > NUM_KERNELS) return ROCPROFILER_ATT_CONTROL_NONE;
if(count.fetch_add(1) > NUM_KERNELS) return ROCPROFILER_THREAD_TRACE_CONTROL_NONE;
assert(userdata && "Dispatch callback passed null!");
dispatch_userdata->ptr = userdata;
return ROCPROFILER_ATT_CONTROL_START_AND_STOP;
return ROCPROFILER_THREAD_TRACE_CONTROL_START_AND_STOP;
}
int
@@ -70,8 +70,8 @@ tool_init(rocprofiler_client_finalize_t /* fini_func */, void* tool_data)
tool_data),
"code object tracing service configure");
std::vector<rocprofiler_att_parameter_t> params{};
params.push_back({ROCPROFILER_ATT_PARAMETER_SERIALIZE_ALL, 1});
std::vector<rocprofiler_thread_trace_parameter_t> params{};
params.push_back({ROCPROFILER_THREAD_TRACE_PARAMETER_SERIALIZE_ALL, 1});
std::vector<rocprofiler_agent_id_t> agents{};
@@ -35,7 +35,7 @@ namespace Single
{
rocprofiler_client_id_t* client_id = nullptr;
rocprofiler_att_control_flags_t
rocprofiler_thread_trace_control_flags_t
dispatch_callback(rocprofiler_agent_id_t /* agent */,
rocprofiler_queue_id_t /* queue_id */,
rocprofiler_async_correlation_id_t /* correlation_id */,
@@ -55,16 +55,16 @@ dispatch_callback(rocprofiler_agent_id_t /* agent */,
{
auto& kernel_name = tool.kernel_id_to_kernel_name.at(kernel_id);
if(kernel_name.find(desired_func_name) == std::string::npos)
return ROCPROFILER_ATT_CONTROL_NONE;
return ROCPROFILER_THREAD_TRACE_CONTROL_NONE;
return ROCPROFILER_ATT_CONTROL_START_AND_STOP;
return ROCPROFILER_THREAD_TRACE_CONTROL_START_AND_STOP;
} catch(...)
{
std::cerr << "Could not find kernel id: " << kernel_id << std::endl;
}
C_API_END
return ROCPROFILER_ATT_CONTROL_NONE;
return ROCPROFILER_THREAD_TRACE_CONTROL_NONE;
}
int