gh-116968: Reimplement Tier 2 counters (#117144)

Introduce a unified 16-bit backoff counter type (``_Py_BackoffCounter``),
shared between the Tier 1 adaptive specializer and the Tier 2 optimizer. The
API used for adaptive specialization counters is changed but the behavior is
(supposed to be) identical.

The behavior of the Tier 2 counters is changed:
- There are no longer dynamic thresholds (we never varied these).
- All counters now use the same exponential backoff.
- The counter for ``JUMP_BACKWARD`` starts counting down from 16.
- The ``temperature`` in side exits starts counting down from 64.

Include/internal/pycore_code.h

@@ -31,7 +31,7 @@ extern "C" {
 #define CACHE_ENTRIES(cache) (sizeof(cache)/sizeof(_Py_CODEUNIT))
 
 typedef struct {
-    uint16_t counter;
+    _Py_BackoffCounter counter;
     uint16_t module_keys_version;
     uint16_t builtin_keys_version;
     uint16_t index;
@@ -40,44 +40,44 @@ typedef struct {
 #define INLINE_CACHE_ENTRIES_LOAD_GLOBAL CACHE_ENTRIES(_PyLoadGlobalCache)
 
 typedef struct {
-    uint16_t counter;
+    _Py_BackoffCounter counter;
 } _PyBinaryOpCache;
 
 #define INLINE_CACHE_ENTRIES_BINARY_OP CACHE_ENTRIES(_PyBinaryOpCache)
 
 typedef struct {
-    uint16_t counter;
+    _Py_BackoffCounter counter;
 } _PyUnpackSequenceCache;
 
 #define INLINE_CACHE_ENTRIES_UNPACK_SEQUENCE \
     CACHE_ENTRIES(_PyUnpackSequenceCache)
 
 typedef struct {
-    uint16_t counter;
+    _Py_BackoffCounter counter;
 } _PyCompareOpCache;
 
 #define INLINE_CACHE_ENTRIES_COMPARE_OP CACHE_ENTRIES(_PyCompareOpCache)
 
 typedef struct {
-    uint16_t counter;
+    _Py_BackoffCounter counter;
 } _PyBinarySubscrCache;
 
 #define INLINE_CACHE_ENTRIES_BINARY_SUBSCR CACHE_ENTRIES(_PyBinarySubscrCache)
 
 typedef struct {
-    uint16_t counter;
+    _Py_BackoffCounter counter;
 } _PySuperAttrCache;
 
 #define INLINE_CACHE_ENTRIES_LOAD_SUPER_ATTR CACHE_ENTRIES(_PySuperAttrCache)
 
 typedef struct {
-    uint16_t counter;
+    _Py_BackoffCounter counter;
     uint16_t version[2];
     uint16_t index;
 } _PyAttrCache;
 
 typedef struct {
-    uint16_t counter;
+    _Py_BackoffCounter counter;
     uint16_t type_version[2];
     union {
         uint16_t keys_version[2];
@@ -93,39 +93,39 @@ typedef struct {
 #define INLINE_CACHE_ENTRIES_STORE_ATTR CACHE_ENTRIES(_PyAttrCache)
 
 typedef struct {
-    uint16_t counter;
+    _Py_BackoffCounter counter;
     uint16_t func_version[2];
 } _PyCallCache;
 
 #define INLINE_CACHE_ENTRIES_CALL CACHE_ENTRIES(_PyCallCache)
 
 typedef struct {
-    uint16_t counter;
+    _Py_BackoffCounter counter;
 } _PyStoreSubscrCache;
 
 #define INLINE_CACHE_ENTRIES_STORE_SUBSCR CACHE_ENTRIES(_PyStoreSubscrCache)
 
 typedef struct {
-    uint16_t counter;
+    _Py_BackoffCounter counter;
 } _PyForIterCache;
 
 #define INLINE_CACHE_ENTRIES_FOR_ITER CACHE_ENTRIES(_PyForIterCache)
 
 typedef struct {
-    uint16_t counter;
+    _Py_BackoffCounter counter;
 } _PySendCache;
 
 #define INLINE_CACHE_ENTRIES_SEND CACHE_ENTRIES(_PySendCache)
 
 typedef struct {
-    uint16_t counter;
+    _Py_BackoffCounter counter;
     uint16_t version[2];
 } _PyToBoolCache;
 
 #define INLINE_CACHE_ENTRIES_TO_BOOL CACHE_ENTRIES(_PyToBoolCache)
 
 typedef struct {
-    uint16_t counter;
+    _Py_BackoffCounter counter;
 } _PyContainsOpCache;
 
 #define INLINE_CACHE_ENTRIES_CONTAINS_OP CACHE_ENTRIES(_PyContainsOpCache)
@@ -451,18 +451,14 @@ write_location_entry_start(uint8_t *ptr, int code, int length)
 
 /** Counters
  * The first 16-bit value in each inline cache is a counter.
- * When counting misses, the counter is treated as a simple unsigned value.
  *
  * When counting executions until the next specialization attempt,
  * exponential backoff is used to reduce the number of specialization failures.
- * The high 12 bits store the counter, the low 4 bits store the backoff exponent.
- * On a specialization failure, the backoff exponent is incremented and the
- * counter set to (2**backoff - 1).
- * Backoff == 6 -> starting counter == 63, backoff == 10 -> starting counter == 1023.
+ * See pycore_backoff.h for more details.
+ * On a specialization failure, the backoff counter is restarted.
  */
 
-/* With a 16-bit counter, we have 12 bits for the counter value, and 4 bits for the backoff */
-#define ADAPTIVE_BACKOFF_BITS 4
+#include "pycore_backoff.h"
 
 // A value of 1 means that we attempt to specialize the *second* time each
 // instruction is executed. Executing twice is a much better indicator of
@@ -480,36 +476,30 @@ write_location_entry_start(uint8_t *ptr, int code, int length)
 #define ADAPTIVE_COOLDOWN_VALUE 52
 #define ADAPTIVE_COOLDOWN_BACKOFF 0
 
-#define MAX_BACKOFF_VALUE (16 - ADAPTIVE_BACKOFF_BITS)
+// Can't assert this in pycore_backoff.h because of header order dependencies
+static_assert(COLD_EXIT_INITIAL_VALUE > ADAPTIVE_COOLDOWN_VALUE,
+    "Cold exit value should be larger than adaptive cooldown value");
 
-
-static inline uint16_t
+static inline _Py_BackoffCounter
 adaptive_counter_bits(uint16_t value, uint16_t backoff) {
-    return ((value << ADAPTIVE_BACKOFF_BITS)
-            | (backoff & ((1 << ADAPTIVE_BACKOFF_BITS) - 1)));
+    return make_backoff_counter(value, backoff);
 }
 
-static inline uint16_t
+static inline _Py_BackoffCounter
 adaptive_counter_warmup(void) {
     return adaptive_counter_bits(ADAPTIVE_WARMUP_VALUE,
                                  ADAPTIVE_WARMUP_BACKOFF);
 }
 
-static inline uint16_t
+static inline _Py_BackoffCounter
 adaptive_counter_cooldown(void) {
     return adaptive_counter_bits(ADAPTIVE_COOLDOWN_VALUE,
                                  ADAPTIVE_COOLDOWN_BACKOFF);
 }
 
-static inline uint16_t
-adaptive_counter_backoff(uint16_t counter) {
-    uint16_t backoff = counter & ((1 << ADAPTIVE_BACKOFF_BITS) - 1);
-    backoff++;
-    if (backoff > MAX_BACKOFF_VALUE) {
-        backoff = MAX_BACKOFF_VALUE;
-    }
-    uint16_t value = (uint16_t)(1 << backoff) - 1;
-    return adaptive_counter_bits(value, backoff);
+static inline _Py_BackoffCounter
+adaptive_counter_backoff(_Py_BackoffCounter counter) {
+    return restart_backoff_counter(counter);
 }