Add MODE=optlinux build mode (#141)

third_party/python/Objects/dictobject.c

@@ -6,6 +6,7 @@
 ╚─────────────────────────────────────────────────────────────────────────────*/
 #include "libc/bits/likely.h"
 #include "libc/calls/calls.h"
+#include "libc/log/countbranch.h"
 #include "libc/runtime/runtime.h"
 #include "libc/sysv/consts/o.h"
 #include "third_party/python/Include/abstract.h"
@@ -1422,10 +1423,10 @@ PyDict_GetItem(PyObject *op, PyObject *key)
     PyThreadState *tstate;
     PyObject **value_addr;
 
-    if (!PyDict_Check(op))
+    if (UNLIKELY(!PyDict_Check(op)))
         return NULL;
-    if (!PyUnicode_CheckExact(key) ||
-        (hash = ((PyASCIIObject *) key)->hash) == -1)
+    if (UNLIKELY(!PyUnicode_CheckExact(key)) ||
+        UNLIKELY((hash = ((PyASCIIObject *) key)->hash) == -1))
     {
         hash = PyObject_Hash(key);
         if (hash == -1) {
@@ -1440,7 +1441,7 @@ PyDict_GetItem(PyObject *op, PyObject *key)
        _PyThreadState_Current and not PyThreadState_GET() because in debug
        mode, the latter complains if tstate is NULL. */
     tstate = _PyThreadState_UncheckedGet();
-    if (tstate != NULL && tstate->curexc_type != NULL) {
+    if (UNLIKELY(tstate != NULL && tstate->curexc_type != NULL)) {
         /* preserve the existing exception */
         PyObject *err_type, *err_value, *err_tb;
         PyErr_Fetch(&err_type, &err_value, &err_tb);

third_party/python/Objects/object.c

@@ -4,6 +4,8 @@
 │ Python 3                                                                     │
 │ https://docs.python.org/3/license.html                                       │
 ╚─────────────────────────────────────────────────────────────────────────────*/
+#include "libc/bits/likely.h"
+#include "libc/log/countbranch.h"
 #include "third_party/python/Include/abstract.h"
 #include "third_party/python/Include/boolobject.h"
 #include "third_party/python/Include/bytearrayobject.h"
@@ -1101,7 +1103,7 @@ _PyObject_GenericGetAttrWithDict(PyObject *obj, PyObject *name, PyObject *dict)
     Py_ssize_t dictoffset;
     PyObject **dictptr;
 
-    if (!PyUnicode_Check(name)){
+    if (UNLIKELY(!PyUnicode_Check(name))){
         PyErr_Format(PyExc_TypeError,
                      "attribute name must be string, not '%.200s'",
                      name->ob_type->tp_name);
@@ -1109,7 +1111,7 @@ _PyObject_GenericGetAttrWithDict(PyObject *obj, PyObject *name, PyObject *dict)
     }
     Py_INCREF(name);
 
-    if (tp->tp_dict == NULL) {
+    if (UNLIKELY(tp->tp_dict == NULL)) {
         if (PyType_Ready(tp) < 0)
             goto done;
     }
@@ -1126,7 +1128,7 @@ _PyObject_GenericGetAttrWithDict(PyObject *obj, PyObject *name, PyObject *dict)
         }
     }
 
-    if (dict == NULL) {
+    if (LIKELY(dict == NULL)) {
         /* Inline _PyObject_GetDictPtr */
         dictoffset = tp->tp_dictoffset;
         if (dictoffset != 0) {
@@ -1148,6 +1150,7 @@ _PyObject_GenericGetAttrWithDict(PyObject *obj, PyObject *name, PyObject *dict)
             dict = *dictptr;
         }
     }
+
     if (dict != NULL) {
         Py_INCREF(dict);
         res = PyDict_GetItem(dict, name);

third_party/python/Objects/typeobject.c

@@ -5,7 +5,9 @@
 │ https://docs.python.org/3/license.html                                       │
 ╚─────────────────────────────────────────────────────────────────────────────*/
 #include "libc/assert.h"
+#include "libc/bits/likely.h"
 #include "libc/fmt/fmt.h"
+#include "libc/log/countbranch.h"
 #include "third_party/python/Include/abstract.h"
 #include "third_party/python/Include/boolobject.h"
 #include "third_party/python/Include/cellobject.h"
@@ -2940,12 +2942,12 @@ _PyType_Lookup(PyTypeObject *type, PyObject *name)
     PyObject *mro, *res, *base, *dict;
     unsigned int h;
 
-    if (MCACHE_CACHEABLE_NAME(name) &&
-        PyType_HasFeature(type, Py_TPFLAGS_VALID_VERSION_TAG)) {
+    if (LIKELY(MCACHE_CACHEABLE_NAME(name)) &&
+        LIKELY(PyType_HasFeature(type, Py_TPFLAGS_VALID_VERSION_TAG))) {
         /* fast path */
         h = MCACHE_HASH_METHOD(type, name);
-        if (method_cache[h].version == type->tp_version_tag &&
-            method_cache[h].name == name) {
+        if (LIKELY(method_cache[h].version == type->tp_version_tag) &&
+            LIKELY(method_cache[h].name == name)) {
 #if MCACHE_STATS
             method_cache_hits++;
 #endif
@@ -2956,9 +2958,9 @@ _PyType_Lookup(PyTypeObject *type, PyObject *name)
     /* Look in tp_dict of types in MRO */
     mro = type->tp_mro;
 
-    if (mro == NULL) {
-        if ((type->tp_flags & Py_TPFLAGS_READYING) == 0 &&
-            PyType_Ready(type) < 0) {
+    if (UNLIKELY(mro == NULL)) {
+        if (UNLIKELY((type->tp_flags & Py_TPFLAGS_READYING) == 0 &&
+                     PyType_Ready(type) < 0)) {
             /* It's not ideal to clear the error condition,
                but this function is documented as not setting
                an exception, and I don't want to change that.

third_party/python/Objects/unicodeobject.c

@@ -10,6 +10,7 @@
 #include "libc/bits/weaken.h"
 #include "libc/errno.h"
 #include "libc/fmt/fmt.h"
+#include "libc/log/countbranch.h"
 #include "libc/str/str.h"
 #include "third_party/python/Include/abstract.h"
 #include "third_party/python/Include/boolobject.h"
@@ -3294,14 +3295,26 @@ PyUnicode_Decode(const char *s,
     Py_buffer info;
     char buflower[11];   /* strlen("iso-8859-1\0") == 11, longest shortcut */
 
-    if (encoding == NULL) {
+    if (UNLIKELY(encoding == NULL)) {
         return PyUnicode_DecodeUTF8Stateful(s, size, errors, NULL);
     }
 
+    /* [jart] faster path based on profiling */
+    if (encoding[0] == 'l' &&
+        encoding[1] == 'a' &&
+        encoding[2] == 't' &&
+        encoding[3] == 'i' &&
+        encoding[4] == 'n' &&
+        (encoding[5] == '1' ||
+         ((encoding[5] == '-' ||
+           encoding[5] == '_') &&
+          encoding[6] == '1'))) {
+        return PyUnicode_DecodeLatin1(s, size, errors);
+    }
+
     /* Shortcuts for common default encodings */
     if (_Py_normalize_encoding(encoding, buflower, sizeof(buflower))) {
         char *lower = buflower;
-
         /* Fast paths */
         if (lower[0] == 'u' && lower[1] == 't' && lower[2] == 'f') {
             lower += 3;
@@ -3309,7 +3322,6 @@ PyUnicode_Decode(const char *s,
                 /* Match "utf8" and "utf_8" */
                 lower++;
             }
-
             if (lower[0] == '8' && lower[1] == 0) {
                 return PyUnicode_DecodeUTF8Stateful(s, size, errors, NULL);
             }