PC sampling: online partial PC sampling decoding (#1004)

* PC sampling: online partial PC sampling decoding

PC sampling service decodes a PC sample partially
by replacing the PC with an id of the loaded code object instance
containing PC and the offset of the PC within that code object instance.

* PC sampling: marker records removed

* PC sampling parser: minor doc update in mock

* PC sampling: introducing rocprofiler_pc_t

* NULL value of the code object id introduced.

* Clarifying documenation related to PC offset.

* PC offset documentation improvement

* PC sampling parser benchmark: Reducing the number of samples to recreate half of performance.

source/include/rocprofiler-sdk/callback_tracing.h

@@ -130,6 +130,11 @@ typedef struct
     };
 } rocprofiler_callback_tracing_code_object_load_data_t;
 
+/**
+ * @brief The NULL value of a code object id. Used when code object is unknown.
+ */
+#define ROCPROFILER_CODE_OBJECT_ID_NONE ROCPROFILER_UINT64_C(0)
+
 /**
  * @brief ROCProfiler Code Object Kernel Symbol Tracer Callback Record.
  *

source/include/rocprofiler-sdk/fwd.h

@@ -411,9 +411,7 @@ typedef enum
 typedef enum
 {
     ROCPROFILER_PC_SAMPLING_RECORD_NONE = 0,
-    ROCPROFILER_PC_SAMPLING_RECORD_SAMPLE,                   ///< ::rocprofiler_pc_sampling_record_t
-    ROCPROFILER_PC_SAMPLING_RECORD_CODE_OBJECT_LOAD_MARKER,  ///< ::rocprofiler_pc_sampling_code_object_load_marker_t
-    ROCPROFILER_PC_SAMPLING_RECORD_CODE_OBJECT_UNLOAD_MARKER,  ///< ::rocprofiler_pc_sampling_code_object_unload_marker_t
+    ROCPROFILER_PC_SAMPLING_RECORD_SAMPLE,  ///< ::rocprofiler_pc_sampling_record_t
     ROCPROFILER_PC_SAMPLING_RECORD_LAST,
 } rocprofiler_pc_sampling_record_kind_t;
 

source/include/rocprofiler-sdk/pc_sampling.h

@@ -239,6 +239,31 @@ typedef struct
     uint32_t arb_state_stall   : 10;
 } rocprofiler_pc_sampling_snapshot_v1_t;
 
+/**
+ * @brief Sampled program counter.
+ */
+typedef struct
+{
+    uint64_t loaded_code_object_id;
+    uint64_t loaded_code_object_offset;
+
+    /// @var loaded_code_object_id
+    /// @brief id of the loaded code object instance that contains sampled PC.
+    /// This fields holds the value ::ROCPROFILER_CODE_OBJECT_ID_NONE
+    /// if the code object cannot be determined
+    /// (e.g., sampled PC belongs to code generated by self modifying code).
+    /// @var loaded_code_object_offset
+    /// @brief If @ref loaded_code_object_id is different than ::ROCPROFILER_CODE_OBJECT_ID_NONE,
+    /// then this field contains the offset of the sampled PC relative to the
+    /// ::rocprofiler_callback_tracing_code_object_load_data_t::load_base
+    /// of the code object instance with @ref loaded_code_object_id.
+    /// To calculate the original virtual address of the sampled PC, one can add the value
+    /// of this field to the ::rocprofiler_callback_tracing_code_object_load_data_t::load_base.
+    /// The value of @ref loaded_code_object_offset matches
+    /// the virtual address of the sampled instruction (PC), only if the
+    /// @ref loaded_code_object_id is equal to the ::ROCPROFILER_CODE_OBJECT_ID_NONE.
+} rocprofiler_pc_t;
+
 // TODO: The definition of this structure might change over time
 // to reduce the space needed to represent a single sample.
 /**
@@ -253,12 +278,12 @@ typedef struct
     uint8_t                             wave_issued : 1;
     uint8_t                             reserved    : 7;  ///< reserved 7 bits, must be zero
     uint32_t                            hw_id;            ///< compute unit identifier
-    uint64_t                     pc;  ///< Program counter of the wave of the moment of interruption
-    uint64_t                     exec_mask;
-    rocprofiler_dim3_t           workgroup_id;  ///< wave coordinates within the workgroup
-    uint32_t                     wave_count;
-    uint64_t                     timestamp;  ///< timestamp when sample is generated
-    rocprofiler_correlation_id_t correlation_id;
+    rocprofiler_pc_t                    pc;  ///< information about sampled program counter
+    uint64_t                            exec_mask;
+    rocprofiler_dim3_t                  workgroup_id;  ///< wave coordinates within the workgroup
+    uint32_t                            wave_count;
+    uint64_t                            timestamp;  ///< timestamp when sample is generated
+    rocprofiler_correlation_id_t        correlation_id;
     rocprofiler_pc_sampling_snapshot_v1_t
              snapshot;   ///< @see ::rocprofiler_pc_sampling_snapshot_v1_t
     uint32_t reserved2;  ///< for future use
@@ -274,42 +299,18 @@ typedef struct
     /// @var wave_count
     /// @brief number of active waves on the CU at the moment of sample generation
     /// @var correlation_id
-    /// @brief correlation id of the API call that initiated kernel launch.
-    /// The interrupted wave is executed as part of the kernel.
+    /// @brief correlation id of the API call that initiated a dispatch of the kernel
+    /// during whose execution the wave was interrupted at @ref pc.
 } rocprofiler_pc_sampling_record_t;
 
-/**
- * @brief Marker representing code object loading event.
- *
- * @see rocprofiler_callback_tracing_code_object_load_data_t
- * for more information
- */
-typedef struct
-{
-    uint64_t size;            ///< Size of this struct
-    uint64_t code_object_id;  /// unique code object identifier
-} rocprofiler_pc_sampling_code_object_load_marker_t;
-
-/**
- * @brief Marker representing code object unloading event.
- *
- * @see rocprofiler_callback_tracing_code_object_load_data_t
- * for more information
- */
-typedef struct
-{
-    uint64_t size;            ///< Size of this struct
-    uint64_t code_object_id;  /// unique code object identifier
-} rocprofiler_pc_sampling_code_object_unload_marker_t;
-
 /** @} */
 
 ROCPROFILER_EXTERN_C_FINI
 
 ROCPROFILER_CXX_CODE(
-    static_assert(sizeof(rocprofiler_pc_sampling_record_t) == 80,
+    static_assert(sizeof(rocprofiler_pc_sampling_record_t) == 88,
                   "Increasing the size of the pc sampling record is not permitted."));
 
 ROCPROFILER_CXX_CODE(static_assert(offsetof(rocprofiler_pc_sampling_record_t, chiplet) == 9 &&
-                                       offsetof(rocprofiler_pc_sampling_record_t, reserved2) == 76,
+                                       offsetof(rocprofiler_pc_sampling_record_t, reserved2) == 84,
                                    "PC sampling record layout changed."));