Add event counter support

Adds support for RSMI event counters. This also includes
"macro" or "pseudo" events, in which an event value is
obtained from RSMI, followed by some post processing before
being displayed in rdci.

Aside from the support of new fields, the main update here
is to introduce an initialization and "shutdown" call for
new fields that will require this.

Also, includes some modifications to the rdci dmon list
command:
* in rdc_field_data.data, added the ability to specify whether
  a field should be hidden or not, by default. This will
  allow us to support many fields, even those that are not
  typically of interest (but sometimes may be), without
  confusing the user or unnecessary clutter.
* added a --list-all option which lists all available field
  including the more obscure fields.

Change-Id: I01dd0edea963c12f82c6e44f893a390711ef3e83

include/rdc_lib/impl/RdcMetricFetcherImpl.h

@@ -29,6 +29,7 @@ THE SOFTWARE.
 #include <queue>
 #include "rdc_lib/RdcMetricFetcher.h"
 #include "rdc_lib/rdc_common.h"
+#include "rocm_smi/rocm_smi.h"
 
 namespace amd {
 namespace rdc {
@@ -41,6 +42,21 @@ struct MetricValue {
     rdc_field_value value;
 };
 
+// This union represents any RSMI handles require initialization and/or
+// shut down. There should only be one instance of this for each raw event
+// used. For example, if a field group includes a pseudo-event and the
+// underlying raw event, then only one FieldRSMIData should be created,
+// and it should be used by both events.
+struct FieldRSMIData {
+    union {
+        rsmi_event_handle_t evt_handle;
+    };
+    union {
+        rsmi_counter_value_t counter_val;
+    };
+    ~FieldRSMIData(){}
+    FieldRSMIData() : evt_handle(0), counter_val{0, 0, 0}{}
+};
 
 //!< The data structure to store the async fetch task
 class RdcMetricFetcherImpl;
@@ -55,7 +71,13 @@ class RdcMetricFetcherImpl: public RdcMetricFetcher {
         rdc_field_t field_id, rdc_field_value* value) override;
     RdcMetricFetcherImpl();
     ~RdcMetricFetcherImpl();
+
+    rdc_status_t acquire_rsmi_handle(RdcFieldKey fk) override;
+    rdc_status_t delete_rsmi_handle(RdcFieldKey fk) override;
+
  private:
+    std::shared_ptr<FieldRSMIData> get_rsmi_data(RdcFieldKey key);
+
     uint64_t now();
     void get_ecc_error(uint32_t gpu_index,
         rdc_field_t field_id, rdc_field_value* value);
@@ -67,6 +89,7 @@ class RdcMetricFetcherImpl: public RdcMetricFetcher {
 
     //!< Async metric retreive
     std::map<RdcFieldKey, MetricValue> async_metrics_;
+    std::map<RdcFieldKey, std::shared_ptr<FieldRSMIData>> rsmi_data_;
     std::queue<MetricTask> updated_tasks_;
     std::mutex task_mutex_;
     std::future<void> updater_;  // keep the future of updater
@@ -74,6 +97,8 @@ class RdcMetricFetcherImpl: public RdcMetricFetcher {
     std::atomic<bool> task_started_;
 };
 
+rdc_status_t Rsmi2RdcError(rsmi_status_t rsmi);
+
 }  // namespace rdc
 }  // namespace amd
 

include/rdc_lib/impl/RdcWatchTableImpl.h

@@ -29,10 +29,12 @@ THE SOFTWARE.
 #include <memory>
 #include <mutex>  // NOLINT
 #include <atomic>
+#include <map>
 #include "rdc_lib/RdcWatchTable.h"
 #include "rdc_lib/RdcGroupSettings.h"
 #include "rdc_lib/RdcCacheManager.h"
 #include "rdc_lib/RdcMetricFetcher.h"
+#include "rocm_smi/rocm_smi.h"
 
 namespace amd {
 namespace rdc {
@@ -51,6 +53,7 @@ struct JobWatchTableEntry {
     std::vector<RdcFieldKey> fields;  //< store fields for faster query
 };
 
+
 class RdcWatchTableImpl : public RdcWatchTable {
  public:
     rdc_status_t rdc_job_start_stats(rdc_gpu_group_t group_id,
@@ -103,6 +106,8 @@ class RdcWatchTableImpl : public RdcWatchTable {
     bool is_job_watch_field(uint32_t gpu_index, rdc_field_t field_id,
                                         std::string& job_id) const;  // NOLINT
 
+    rdc_status_t initialize_rsmi_handles(RdcFieldKey fk);
+
     RdcGroupSettingsPtr group_settings_;
     RdcCacheManagerPtr cache_mgr_;
     RdcMetricFetcherPtr metric_fetcher_;
@@ -122,7 +127,6 @@ class RdcWatchTableImpl : public RdcWatchTable {
 
     //!< The last clean up time
     std::atomic<uint64_t> last_cleanup_time_;
-
     std::mutex watch_mutex_;
 };