GH-109329: Add tier 2 stats (GH-109913)

mdboom · web-flow · commit e561e9805854 · 2023-10-04T14:52:28.000-07:00
diff --git a/Include/cpython/pystats.h b/Include/cpython/pystats.h
@@ -86,10 +86,6 @@ typedef struct _object_stats {
     uint64_t type_cache_dunder_hits;
     uint64_t type_cache_dunder_misses;
     uint64_t type_cache_collisions;
-    uint64_t optimization_attempts;
-    uint64_t optimization_traces_created;
-    uint64_t optimization_traces_executed;
-    uint64_t optimization_uops_executed;
     /* Temporary value used during GC */
     uint64_t object_visits;
 } ObjectStats;
@@ -100,10 +96,34 @@ typedef struct _gc_stats {
     uint64_t objects_collected;
 } GCStats;
 
+typedef struct _uop_stats {
+    uint64_t execution_count;
+} UOpStats;
+
+#define _Py_UOP_HIST_SIZE 32
+
+typedef struct _optimization_stats {
+    uint64_t attempts;
+    uint64_t traces_created;
+    uint64_t traces_executed;
+    uint64_t uops_executed;
+    uint64_t trace_stack_overflow;
+    uint64_t trace_stack_underflow;
+    uint64_t trace_too_long;
+    uint64_t inner_loop;
+    uint64_t recursive_call;
+    UOpStats opcode[512];
+    uint64_t unsupported_opcode[256];
+    uint64_t trace_length_hist[_Py_UOP_HIST_SIZE];
+    uint64_t trace_run_length_hist[_Py_UOP_HIST_SIZE];
+    uint64_t optimized_trace_length_hist[_Py_UOP_HIST_SIZE];
+} OptimizationStats;
+
 typedef struct _stats {
     OpcodeStats opcode_stats[256];
     CallStats call_stats;
     ObjectStats object_stats;
+    OptimizationStats optimization_stats;
     GCStats *gc_stats;
 } PyStats;
 
diff --git a/Include/internal/pycore_code.h b/Include/internal/pycore_code.h
@@ -282,6 +282,17 @@ extern int _PyStaticCode_Init(PyCodeObject *co);
 #define EVAL_CALL_STAT_INC_IF_FUNCTION(name, callable) \
     do { if (_Py_stats && PyFunction_Check(callable)) _Py_stats->call_stats.eval_calls[name]++; } while (0)
 #define GC_STAT_ADD(gen, name, n) do { if (_Py_stats) _Py_stats->gc_stats[(gen)].name += (n); } while (0)
+#define OPT_STAT_INC(name) do { if (_Py_stats) _Py_stats->optimization_stats.name++; } while (0)
+#define UOP_EXE_INC(opname) do { if (_Py_stats) _Py_stats->optimization_stats.opcode[opname].execution_count++; } while (0)
+#define OPT_UNSUPPORTED_OPCODE(opname) do { if (_Py_stats) _Py_stats->optimization_stats.unsupported_opcode[opname]++; } while (0)
+#define OPT_HIST(length, name) \
+    do { \
+        if (_Py_stats) { \
+            int bucket = _Py_bit_length(length >= 1 ? length - 1 : 0); \
+            bucket = (bucket >= _Py_UOP_HIST_SIZE) ? _Py_UOP_HIST_SIZE - 1 : bucket; \
+            _Py_stats->optimization_stats.name[bucket]++; \
+        } \
+    } while (0)
 
 // Export for '_opcode' shared extension
 PyAPI_FUNC(PyObject*) _Py_GetSpecializationStats(void);
@@ -296,6 +307,10 @@ PyAPI_FUNC(PyObject*) _Py_GetSpecializationStats(void);
 #define EVAL_CALL_STAT_INC(name) ((void)0)
 #define EVAL_CALL_STAT_INC_IF_FUNCTION(name, callable) ((void)0)
 #define GC_STAT_ADD(gen, name, n) ((void)0)
+#define OPT_STAT_INC(name) ((void)0)
+#define UOP_EXE_INC(opname) ((void)0)
+#define OPT_UNSUPPORTED_OPCODE(opname) ((void)0)
+#define OPT_HIST(length, name) ((void)0)
 #endif  // !Py_STATS
 
 // Utility functions for reading/writing 32/64-bit values in the inline caches.
diff --git a/Python/bytecodes.c b/Python/bytecodes.c
@@ -2278,7 +2278,7 @@ dummy_func(
                 // Double-check that the opcode isn't instrumented or something:
                 here->op.code == JUMP_BACKWARD)
             {
-                OBJECT_STAT_INC(optimization_attempts);
+                OPT_STAT_INC(attempts);
                 int optimized = _PyOptimizer_BackEdge(frame, here, next_instr, stack_pointer);
                 ERROR_IF(optimized < 0, error);
                 if (optimized) {
@@ -3965,6 +3965,7 @@ dummy_func(
             frame->prev_instr--;  // Back up to just before destination
             _PyFrame_SetStackPointer(frame, stack_pointer);
             Py_DECREF(self);
+            OPT_HIST(trace_uop_execution_counter, trace_run_length_hist);
             return frame;
         }
 
diff --git a/Python/executor.c b/Python/executor.c
@@ -2,6 +2,7 @@
 
 #include "opcode.h"
 
+#include "pycore_bitutils.h"
 #include "pycore_call.h"
 #include "pycore_ceval.h"
 #include "pycore_dict.h"
@@ -30,6 +31,16 @@
         goto deoptimize;         \
     }
 
+#ifdef Py_STATS
+// Disable these macros that apply to Tier 1 stats when we are in Tier 2
+#undef STAT_INC
+#define STAT_INC(opname, name) ((void)0)
+#undef STAT_DEC
+#define STAT_DEC(opname, name) ((void)0)
+#undef CALL_STAT_INC
+#define CALL_STAT_INC(name) ((void)0)
+#endif
+
 #undef ENABLE_SPECIALIZATION
 #define ENABLE_SPECIALIZATION 0
 
@@ -62,12 +73,15 @@ _PyUopExecute(_PyExecutorObject *executor, _PyInterpreterFrame *frame, PyObject
 
     CHECK_EVAL_BREAKER();
 
-    OBJECT_STAT_INC(optimization_traces_executed);
+    OPT_STAT_INC(traces_executed);
     _Py_CODEUNIT *ip_offset = (_Py_CODEUNIT *)_PyFrame_GetCode(frame)->co_code_adaptive;
     int pc = 0;
     int opcode;
     int oparg;
     uint64_t operand;
+#ifdef Py_STATS
+    uint64_t trace_uop_execution_counter = 0;
+#endif
 
     for (;;) {
         opcode = self->trace[pc].opcode;
@@ -81,7 +95,12 @@ _PyUopExecute(_PyExecutorObject *executor, _PyInterpreterFrame *frame, PyObject
                 operand,
                 (int)(stack_pointer - _PyFrame_Stackbase(frame)));
         pc++;
-        OBJECT_STAT_INC(optimization_uops_executed);
+        OPT_STAT_INC(uops_executed);
+        UOP_EXE_INC(opcode);
+#ifdef Py_STATS
+        trace_uop_execution_counter++;
+#endif
+
         switch (opcode) {
 
 #include "executor_cases.c.h"
@@ -114,6 +133,7 @@ _PyUopExecute(_PyExecutorObject *executor, _PyInterpreterFrame *frame, PyObject
     // On ERROR_IF we return NULL as the frame.
     // The caller recovers the frame from tstate->current_frame.
     DPRINTF(2, "Error: [Opcode %d, operand %" PRIu64 "]\n", opcode, operand);
+    OPT_HIST(trace_uop_execution_counter, trace_run_length_hist);
     _PyFrame_SetStackPointer(frame, stack_pointer);
     Py_DECREF(self);
     return NULL;
@@ -122,6 +142,7 @@ _PyUopExecute(_PyExecutorObject *executor, _PyInterpreterFrame *frame, PyObject
     // On DEOPT_IF we just repeat the last instruction.
     // This presumes nothing was popped from the stack (nor pushed).
     DPRINTF(2, "DEOPT: [Opcode %d, operand %" PRIu64 "]\n", opcode, operand);
+    OPT_HIST(trace_uop_execution_counter, trace_run_length_hist);
     frame->prev_instr--;  // Back up to just before destination
     _PyFrame_SetStackPointer(frame, stack_pointer);
     Py_DECREF(self);
diff --git a/Python/executor_cases.c.h b/Python/executor_cases.c.h
diff --git a/Python/generated_cases.c.h b/Python/generated_cases.c.h
diff --git a/Python/optimizer.c b/Python/optimizer.c
@@ -461,6 +461,7 @@ translate_bytecode_to_trace(
     if (trace_length + (n) > max_length) { \
         DPRINTF(2, "No room for %s (need %d, got %d)\n", \
                 (opname), (n), max_length - trace_length); \
+        OPT_STAT_INC(trace_too_long); \
         goto done; \
     } \
     reserved = (n);  // Keep ADD_TO_TRACE / ADD_TO_STUB honest
@@ -472,6 +473,7 @@ translate_bytecode_to_trace(
 #define TRACE_STACK_PUSH() \
     if (trace_stack_depth >= TRACE_STACK_SIZE) { \
         DPRINTF(2, "Trace stack overflow\n"); \
+        OPT_STAT_INC(trace_stack_overflow); \
         ADD_TO_TRACE(_SET_IP, 0, 0); \
         goto done; \
     } \
@@ -572,6 +574,7 @@ translate_bytecode_to_trace(
                     ADD_TO_TRACE(_JUMP_TO_TOP, 0, 0);
                 }
                 else {
+                    OPT_STAT_INC(inner_loop);
                     DPRINTF(2, "JUMP_BACKWARD not to top ends trace\n");
                 }
                 goto done;
@@ -638,7 +641,9 @@ translate_bytecode_to_trace(
                         // LOAD_CONST + _POP_FRAME.
                         if (trace_stack_depth == 0) {
                             DPRINTF(2, "Trace stack underflow\n");
-                            goto done;}
+                            OPT_STAT_INC(trace_stack_underflow);
+                            goto done;
+                        }
                     }
                     uint32_t orig_oparg = oparg;  // For OPARG_TOP/BOTTOM
                     for (int i = 0; i < nuops; i++) {
@@ -713,6 +718,7 @@ translate_bytecode_to_trace(
                                             PyUnicode_AsUTF8(new_code->co_qualname),
                                             PyUnicode_AsUTF8(new_code->co_filename),
                                             new_code->co_firstlineno);
+                                    OPT_STAT_INC(recursive_call);
                                     ADD_TO_TRACE(_SET_IP, 0, 0);
                                     goto done;
                                 }
@@ -744,6 +750,7 @@ translate_bytecode_to_trace(
                     break;
                 }
                 DPRINTF(2, "Unsupported opcode %s\n", uop_name(opcode));
+                OPT_UNSUPPORTED_OPCODE(opcode);
                 goto done;  // Break out of loop
             }  // End default
 
@@ -891,7 +898,8 @@ uop_optimize(
         // Error or nothing translated
         return trace_length;
     }
-    OBJECT_STAT_INC(optimization_traces_created);
+    OPT_HIST(trace_length, trace_length_hist);
+    OPT_STAT_INC(traces_created);
     char *uop_optimize = Py_GETENV("PYTHONUOPSOPTIMIZE");
     if (uop_optimize != NULL && *uop_optimize > '0') {
         trace_length = _Py_uop_analyze_and_optimize(code, trace, trace_length, curr_stackentries);
@@ -901,6 +909,7 @@ uop_optimize(
     if (executor == NULL) {
         return -1;
     }
+    OPT_HIST(trace_length, optimized_trace_length_hist);
     executor->base.execute = _PyUopExecute;
     memcpy(executor->trace, trace, trace_length * sizeof(_PyUOpInstruction));
     *exec_ptr = (_PyExecutorObject *)executor;
diff --git a/Python/specialize.c b/Python/specialize.c
@@ -199,10 +199,6 @@ print_object_stats(FILE *out, ObjectStats *stats)
     fprintf(out, "Object method cache collisions: %" PRIu64 "\n", stats->type_cache_collisions);
     fprintf(out, "Object method cache dunder hits: %" PRIu64 "\n", stats->type_cache_dunder_hits);
     fprintf(out, "Object method cache dunder misses: %" PRIu64 "\n", stats->type_cache_dunder_misses);
-    fprintf(out, "Optimization attempts: %" PRIu64 "\n", stats->optimization_attempts);
-    fprintf(out, "Optimization traces created: %" PRIu64 "\n", stats->optimization_traces_created);
-    fprintf(out, "Optimization traces executed: %" PRIu64 "\n", stats->optimization_traces_executed);
-    fprintf(out, "Optimization uops executed: %" PRIu64 "\n", stats->optimization_uops_executed);
 }
 
 static void
@@ -215,13 +211,63 @@ print_gc_stats(FILE *out, GCStats *stats)
     }
 }
 
+static void
+print_histogram(FILE *out, const char *name, uint64_t hist[_Py_UOP_HIST_SIZE])
+{
+    for (int i = 0; i < _Py_UOP_HIST_SIZE; i++) {
+        fprintf(out, "%s[%" PRIu64"]: %" PRIu64 "\n", name, (uint64_t)1 << i, hist[i]);
+    }
+}
+
+static void
+print_optimization_stats(FILE *out, OptimizationStats *stats)
+{
+    fprintf(out, "Optimization attempts: %" PRIu64 "\n", stats->attempts);
+    fprintf(out, "Optimization traces created: %" PRIu64 "\n", stats->traces_created);
+    fprintf(out, "Optimization traces executed: %" PRIu64 "\n", stats->traces_executed);
+    fprintf(out, "Optimization uops executed: %" PRIu64 "\n", stats->uops_executed);
+    fprintf(out, "Optimization trace stack overflow: %" PRIu64 "\n", stats->trace_stack_overflow);
+    fprintf(out, "Optimization trace stack underflow: %" PRIu64 "\n", stats->trace_stack_underflow);
+    fprintf(out, "Optimization trace too long: %" PRIu64 "\n", stats->trace_too_long);
+    fprintf(out, "Optimization inner loop: %" PRIu64 "\n", stats->inner_loop);
+    fprintf(out, "Optimization recursive call: %" PRIu64 "\n", stats->recursive_call);
+
+    print_histogram(out, "Trace length", stats->trace_length_hist);
+    print_histogram(out, "Trace run length", stats->trace_run_length_hist);
+    print_histogram(out, "Optimized trace length", stats->optimized_trace_length_hist);
+
+    const char* const* names;
+    for (int i = 0; i < 512; i++) {
+        if (i < 256) {
+            names = _PyOpcode_OpName;
+        } else {
+            names = _PyOpcode_uop_name;
+        }
+        if (stats->opcode[i].execution_count) {
+            fprintf(out, "uops[%s].execution_count : %" PRIu64 "\n", names[i], stats->opcode[i].execution_count);
+        }
+    }
+
+    for (int i = 0; i < 256; i++) {
+        if (stats->unsupported_opcode[i]) {
+            fprintf(
+                out,
+                "unsupported_opcode[%s].count : %" PRIu64 "\n",
+                _PyOpcode_OpName[i],
+                stats->unsupported_opcode[i]
+            );
+        }
+    }
+}
+
 static void
 print_stats(FILE *out, PyStats *stats)
 {
     print_spec_stats(out, stats->opcode_stats);
     print_call_stats(out, &stats->call_stats);
     print_object_stats(out, &stats->object_stats);
     print_gc_stats(out, stats->gc_stats);
+    print_optimization_stats(out, &stats->optimization_stats);
 }
 
 void
diff --git a/Tools/scripts/summarize_stats.py b/Tools/scripts/summarize_stats.py
