From c07091fb4219112a1726ac21a57b473f69fcb889 Mon Sep 17 00:00:00 2001 From: Victor Stinner Date: Tue, 5 May 2020 17:42:12 +0200 Subject: [PATCH] bpo-40521: Disable Unicode caches in isolated subinterpreters When Python is build in the experimental isolated subinterpreters mode, disable Unicode singletons and Unicode interned strings since they are shared by all interpreters. Temporary workaround until these caches are made per-interpreter. --- Objects/typeobject.c | 16 +++++++++ Objects/unicodeobject.c | 78 +++++++++++++++++++++++++++++++++-------- 2 files changed, 79 insertions(+), 15 deletions(-) diff --git a/Objects/typeobject.c b/Objects/typeobject.c index db0ae970090ba9..1565b90898605e 100644 --- a/Objects/typeobject.c +++ b/Objects/typeobject.c @@ -56,6 +56,11 @@ static size_t method_cache_misses = 0; static size_t method_cache_collisions = 0; #endif +/* bpo-40521: Interned strings are shared by all subinterpreters */ +#ifndef EXPERIMENTAL_ISOLATED_SUBINTERPRETERS +# define INTERN_NAME_STRINGS +#endif + /* alphabetical order */ _Py_IDENTIFIER(__abstractmethods__); _Py_IDENTIFIER(__class__); @@ -3418,6 +3423,7 @@ type_setattro(PyTypeObject *type, PyObject *name, PyObject *value) if (name == NULL) return -1; } +#ifdef INTERN_NAME_STRINGS if (!PyUnicode_CHECK_INTERNED(name)) { PyUnicode_InternInPlace(&name); if (!PyUnicode_CHECK_INTERNED(name)) { @@ -3427,6 +3433,7 @@ type_setattro(PyTypeObject *type, PyObject *name, PyObject *value) return -1; } } +#endif } else { /* Will fail in _PyObject_GenericSetAttrWithDict. */ @@ -7531,10 +7538,17 @@ _PyTypes_InitSlotDefs(void) for (slotdef *p = slotdefs; p->name; p++) { /* Slots must be ordered by their offset in the PyHeapTypeObject. */ assert(!p[1].name || p->offset <= p[1].offset); +#ifdef INTERN_NAME_STRINGS p->name_strobj = PyUnicode_InternFromString(p->name); if (!p->name_strobj || !PyUnicode_CHECK_INTERNED(p->name_strobj)) { return _PyStatus_NO_MEMORY(); } +#else + p->name_strobj = PyUnicode_FromString(p->name); + if (!p->name_strobj) { + return _PyStatus_NO_MEMORY(); + } +#endif } slotdefs_initialized = 1; return _PyStatus_OK(); @@ -7559,7 +7573,9 @@ update_slot(PyTypeObject *type, PyObject *name) int offset; assert(PyUnicode_CheckExact(name)); +#ifdef INTERN_NAME_STRINGS assert(PyUnicode_CHECK_INTERNED(name)); +#endif assert(slotdefs_initialized); pp = ptrs; diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c index aba7407533c4ed..18b9458721de18 100644 --- a/Objects/unicodeobject.c +++ b/Objects/unicodeobject.c @@ -198,6 +198,11 @@ extern "C" { # define OVERALLOCATE_FACTOR 4 #endif +/* bpo-40521: Interned strings are shared by all interpreters. */ +#ifndef EXPERIMENTAL_ISOLATED_SUBINTERPRETERS +# define INTERNED_STRINGS +#endif + /* This dictionary holds all interned unicode strings. Note that references to strings in this dictionary are *not* counted in the string's ob_refcnt. When the interned string reaches a refcnt of 0 the string deallocation @@ -206,7 +211,9 @@ extern "C" { Another way to look at this is that to say that the actual reference count of a string is: s->ob_refcnt + (s->state ? 2 : 0) */ +#ifdef INTERNED_STRINGS static PyObject *interned = NULL; +#endif /* The empty Unicode object is shared to improve performance. */ static PyObject *unicode_empty = NULL; @@ -281,9 +288,16 @@ unicode_decode_utf8(const char *s, Py_ssize_t size, /* List of static strings. */ static _Py_Identifier *static_strings = NULL; +/* bpo-40521: Latin1 singletons are shared by all interpreters. */ +#ifndef EXPERIMENTAL_ISOLATED_SUBINTERPRETERS +# define LATIN1_SINGLETONS +#endif + +#ifdef LATIN1_SINGLETONS /* Single character Unicode strings in the Latin-1 range are being shared as well. */ static PyObject *unicode_latin1[256] = {NULL}; +#endif /* Fast detection of the most frequent whitespace characters */ const unsigned char _Py_ascii_whitespace[] = { @@ -662,6 +676,7 @@ unicode_result_ready(PyObject *unicode) return unicode_empty; } +#ifdef LATIN1_SINGLETONS if (length == 1) { const void *data = PyUnicode_DATA(unicode); int kind = PyUnicode_KIND(unicode); @@ -683,6 +698,7 @@ unicode_result_ready(PyObject *unicode) } } } +#endif assert(_PyUnicode_CheckConsistency(unicode, 1)); return unicode; @@ -1913,10 +1929,12 @@ unicode_dealloc(PyObject *unicode) case SSTATE_INTERNED_MORTAL: /* revive dead object temporarily for DelItem */ Py_SET_REFCNT(unicode, 3); +#ifdef INTERNED_STRINGS if (PyDict_DelItem(interned, unicode) != 0) { _PyErr_WriteUnraisableMsg("deletion of interned string failed", NULL); } +#endif break; case SSTATE_INTERNED_IMMORTAL: @@ -1944,15 +1962,18 @@ unicode_dealloc(PyObject *unicode) static int unicode_is_singleton(PyObject *unicode) { - PyASCIIObject *ascii = (PyASCIIObject *)unicode; - if (unicode == unicode_empty) + if (unicode == unicode_empty) { return 1; + } +#ifdef LATIN1_SINGLETONS + PyASCIIObject *ascii = (PyASCIIObject *)unicode; if (ascii->state.kind != PyUnicode_WCHAR_KIND && ascii->length == 1) { Py_UCS4 ch = PyUnicode_READ_CHAR(unicode, 0); if (ch < 256 && unicode_latin1[ch] == unicode) return 1; } +#endif return 0; } #endif @@ -2094,16 +2115,28 @@ unicode_write_cstr(PyObject *unicode, Py_ssize_t index, static PyObject* get_latin1_char(unsigned char ch) { - PyObject *unicode = unicode_latin1[ch]; + PyObject *unicode; + +#ifdef LATIN1_SINGLETONS + unicode = unicode_latin1[ch]; + if (unicode) { + Py_INCREF(unicode); + return unicode; + } +#endif + + unicode = PyUnicode_New(1, ch); if (!unicode) { - unicode = PyUnicode_New(1, ch); - if (!unicode) - return NULL; - PyUnicode_1BYTE_DATA(unicode)[0] = ch; - assert(_PyUnicode_CheckConsistency(unicode, 1)); - unicode_latin1[ch] = unicode; + return NULL; } + + PyUnicode_1BYTE_DATA(unicode)[0] = ch; + assert(_PyUnicode_CheckConsistency(unicode, 1)); + +#ifdef LATIN1_SINGLETONS Py_INCREF(unicode); + unicode_latin1[ch] = unicode; +#endif return unicode; } @@ -11270,7 +11303,6 @@ int _PyUnicode_EqualToASCIIId(PyObject *left, _Py_Identifier *right) { PyObject *right_uni; - Py_hash_t hash; assert(_PyUnicode_CHECK(left)); assert(right->string); @@ -11302,10 +11334,12 @@ _PyUnicode_EqualToASCIIId(PyObject *left, _Py_Identifier *right) if (PyUnicode_CHECK_INTERNED(left)) return 0; +#ifdef INTERNED_STRINGS assert(_PyUnicode_HASH(right_uni) != -1); - hash = _PyUnicode_HASH(left); + Py_hash_t hash = _PyUnicode_HASH(left); if (hash != -1 && hash != _PyUnicode_HASH(right_uni)) return 0; +#endif return unicode_compare_eq(left, right_uni); } @@ -15487,20 +15521,26 @@ void PyUnicode_InternInPlace(PyObject **p) { PyObject *s = *p; - PyObject *t; #ifdef Py_DEBUG assert(s != NULL); assert(_PyUnicode_CHECK(s)); #else - if (s == NULL || !PyUnicode_Check(s)) + if (s == NULL || !PyUnicode_Check(s)) { return; + } #endif + /* If it's a subclass, we don't really know what putting it in the interned dict might do. */ - if (!PyUnicode_CheckExact(s)) + if (!PyUnicode_CheckExact(s)) { return; - if (PyUnicode_CHECK_INTERNED(s)) + } + + if (PyUnicode_CHECK_INTERNED(s)) { return; + } + +#ifdef INTERNED_STRINGS if (interned == NULL) { interned = PyDict_New(); if (interned == NULL) { @@ -15508,22 +15548,28 @@ PyUnicode_InternInPlace(PyObject **p) return; } } + + PyObject *t; Py_ALLOW_RECURSION t = PyDict_SetDefault(interned, s, s); Py_END_ALLOW_RECURSION + if (t == NULL) { PyErr_Clear(); return; } + if (t != s) { Py_INCREF(t); Py_SETREF(*p, t); return; } + /* The two references in interned are not counted by refcnt. The deallocator will take care of this */ Py_SET_REFCNT(s, Py_REFCNT(s) - 2); _PyUnicode_STATE(s).interned = SSTATE_INTERNED_MORTAL; +#endif } void @@ -16109,9 +16155,11 @@ _PyUnicode_Fini(PyThreadState *tstate) Py_CLEAR(unicode_empty); +#ifdef LATIN1_SINGLETONS for (Py_ssize_t i = 0; i < 256; i++) { Py_CLEAR(unicode_latin1[i]); } +#endif _PyUnicode_ClearStaticStrings(); }