Skip to content

bpo-47084: Clear cached representations on finalization #32032

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
vstinner merged 6 commits into from
Mar 22, 2022
Merged

bpo-47084: Clear cached representations on finalization #32032

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
4b166f0
Clear cached representations on finalization
jkloth Mar 21, 2022
7d66a24
Merge branch 'main' of https://github.com/python/cpython into issue47084
jkloth Mar 21, 2022
7a4b788
Also reset representation lengths on finalization
jkloth Mar 21, 2022
08c0999
Merge branch 'main' of https://github.com/python/cpython into issue47084
jkloth Mar 21, 2022
55163a2
Use TestCase.subTest() for per-iteration testing
jkloth Mar 21, 2022
72f50c3
Refactor dealloc into single function.
jkloth Mar 21, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions Include/internal/pycore_unicodeobject.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,6 +18,7 @@ extern PyStatus _PyUnicode_InitGlobalObjects(PyInterpreterState *);
extern PyStatus _PyUnicode_InitTypes(PyInterpreterState *);
extern void _PyUnicode_Fini(PyInterpreterState *);
extern void _PyUnicode_FiniTypes(PyInterpreterState *);
extern void _PyStaticUnicode_Dealloc(PyObject *);


/* other API */
Expand Down
9 changes: 9 additions & 0 deletions Lib/__hello__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,14 @@
initialized = True

class TestFrozenUtf8_1:
"""\u00b6"""

class TestFrozenUtf8_2:
"""\u03c0"""

class TestFrozenUtf8_4:
"""\U0001f600"""

def main():
print("Hello world!")

Expand Down
40 changes: 22 additions & 18 deletions Lib/test/test_embed.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1645,24 +1645,28 @@ def test_frozenmain(self):
'-X showrefcount requires a Python debug build')
def test_no_memleak(self):
# bpo-1635741: Python must release all memory at exit
cmd = [sys.executable, "-I", "-X", "showrefcount", "-c", "pass"]
proc = subprocess.run(cmd,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT,
text=True)
self.assertEqual(proc.returncode, 0)
out = proc.stdout.rstrip()
match = re.match(r'^\[(-?\d+) refs, (-?\d+) blocks\]', out)
if not match:
self.fail(f"unexpected output: {out!a}")
refs = int(match.group(1))
blocks = int(match.group(2))
self.assertEqual(refs, 0, out)
if not MS_WINDOWS:
self.assertEqual(blocks, 0, out)
else:
# bpo-46857: on Windows, Python still leaks 1 memory block at exit
self.assertIn(blocks, (0, 1), out)
tests = (
('off', 'pass'),
('on', 'pass'),
('off', 'import __hello__'),
('on', 'import __hello__'),
)
for flag, stmt in tests:

@vstinner vstinner Mar 21, 2022

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can you please use with self.subTest(flag=flag, stmt=stmt):? Otherwise, it will be more complicated to debug if a test fails. With that, you can omit f"{xopt}, stmt='{stmt}' -> {out}") below.

@jkloth jkloth Mar 21, 2022

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Didn't know about TestCase.subTest(), thanks.

xopt = f"frozen_modules={flag}"
cmd = [sys.executable, "-I", "-X", "showrefcount", "-X", xopt, "-c", stmt]
proc = subprocess.run(cmd,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT,
text=True)
self.assertEqual(proc.returncode, 0)
out = proc.stdout.rstrip()
match = re.match(r'^\[(-?\d+) refs, (-?\d+) blocks\]', out)
if not match:
self.fail(f"unexpected output: {out!a}")
refs = int(match.group(1))
blocks = int(match.group(2))
self.assertEqual(refs, 0, f"{xopt}, stmt='{stmt}' -> {out}")
self.assertEqual(blocks, 0, f"{xopt}, stmt='{stmt}' -> {out}")


class StdPrinterTests(EmbeddingTestsMixin, unittest.TestCase):
Expand Down
52 changes: 52 additions & 0 deletions Objects/unicodeobject.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16070,6 +16070,58 @@ _PyUnicode_Fini(PyInterpreterState *interp)
_PyUnicode_FiniEncodings(&state->fs_codec);

unicode_clear_identifiers(state);

// Clear the single character singletons
for (int i = 0; i < 128; i++) {
PyASCIIObject *ascii = (PyASCIIObject *)& _Py_SINGLETON(strings).ascii[i];
if (ascii->wstr != NULL) {
PyObject_FREE(ascii->wstr);
ascii->wstr = NULL;
}
}
for (int i = 0; i < 128; i++) {
PyASCIIObject *ascii = (PyASCIIObject *)&_Py_SINGLETON(strings).latin1[i];
PyCompactUnicodeObject *compact = (PyCompactUnicodeObject *)ascii;
if (ascii->wstr != NULL) {
PyObject_FREE(ascii->wstr);
ascii->wstr = NULL;
compact->wstr_length = 0;
}
if (compact->utf8 != NULL) {
PyObject_Free(compact->utf8);
compact->utf8 = NULL;
compact->utf8_length = 0;
}
}
}


void
_PyStaticUnicode_Dealloc(PyObject *op)

@vstinner vstinner Mar 21, 2022

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Would it make sense to share code between this function and unicode_dealloc()? Maybe add an unicode_dealloc_common() function?

@jkloth jkloth Mar 21, 2022

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, there is the duplication of the PyObect_Free() calls, but not really to the level of what I would consider "in common". I did, however, refactor out the duplication of the 1-character singletons and frozen strings deallocation into a standalone static function.

{
PyASCIIObject *ascii = (PyASCIIObject*)op;

assert(ascii->state.compact);

if (ascii->state.ascii) {
if (ascii->wstr) {
PyObject_Free(ascii->wstr);
ascii->wstr = NULL;
}
} else {
PyCompactUnicodeObject *compact = (PyCompactUnicodeObject*)op;
void *data = (void*)(compact + 1);
if (ascii->wstr && ascii->wstr != data) {
PyObject_Free(ascii->wstr);
ascii->wstr = NULL;
compact->wstr_length = 0;
}
if (compact->utf8) {
PyObject_Free(compact->utf8);
compact->utf8 = NULL;
compact->utf8_length = 0;
}
}
}


Expand Down
1 change: 1 addition & 0 deletions Tools/scripts/deepfreeze.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -185,6 +185,7 @@ def generate_unicode(self, name: str, s: str) -> str:
else:
self.write("PyCompactUnicodeObject _compact;")
self.write(f"{datatype} _data[{len(s)+1}];")
self.deallocs.append(f"_PyStaticUnicode_Dealloc((PyObject *)&{name});")
with self.block(f"{name} =", ";"):
if ascii:
with self.block("._ascii =", ","):
Expand Down