Skip to content

bpo-45850: Implement deep-freeze on Windows #29648

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
merged 18 commits into from
Nov 22, 2021
Merged

bpo-45850: Implement deep-freeze on Windows #29648

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
18 commits
Select commit Hold shift + click to select a range
97c8fd6
Deep-freeze for Windows, step 1
gvanrossum Nov 17, 2021
65ee8fe
Support reading marshalled data from a frozen .h file
gvanrossum Nov 18, 2021
de230cb
Add umarshal.py (3.11 marshal.loads() in pure Python)
gvanrossum Nov 18, 2021
87c6cb4
Generate .c files in _freeze_module.vcxproj
gvanrossum Nov 18, 2021
8b46af7
Avoid copying unchanged deep-frozen files
gvanrossum Nov 19, 2021
1bc777f
Properly generate lines in pythoncore.vcxproj; add codecs back
gvanrossum Nov 19, 2021
97187ff
Try not to break UNIX builds by lazy-importing struct
gvanrossum Nov 19, 2021
55b0f50
Merge remote-tracking branch 'origin/main' into windeepfreeze
gvanrossum Nov 19, 2021
8eeaea0
Remove temporary .bat file
gvanrossum Nov 19, 2021
d332f16
Change ntpath back to posixpath in frozen.c
gvanrossum Nov 19, 2021
3e7c3ac
One more nt->posix
gvanrossum Nov 19, 2021
871f28b
📜🤖 Added by blurb_it.
blurb-it[bot] Nov 20, 2021
b52ad63
Fix by @zooba to get better error when no Python 3.10 is found
gvanrossum Nov 20, 2021
aea38e0
Generate integers for 15- and 30-bit digits
gvanrossum Nov 20, 2021
a7415ed
Give generated .c files a "df." prefix
gvanrossum Nov 20, 2021
f423c3b
Remove unused imports from deepfreeze.py
gvanrossum Nov 20, 2021
a3f60fc
Add a quick script to time startup
gvanrossum Nov 21, 2021
a69895e
Update blurb (reminder that 3.10 must be installed to build)
gvanrossum Nov 22, 2021
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
2 changes: 2 additions & 0 deletions Misc/NEWS.d/next/Windows/2021-11-20-00-06-59.bpo-45850.q9lofz.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,2 @@
Implement changes to build with deep-frozen modules on Windows.
Note that we now require Python 3.10 as the "bootstrap" or "host" Python.
58 changes: 55 additions & 3 deletions PCbuild/_freeze_module.vcxproj
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -234,119 +234,171 @@
<ModName>importlib._bootstrap</ModName>
<IntFile>$(IntDir)importlib._bootstrap.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\importlib._bootstrap.h</OutFile>
<DeepIntFile>$(IntDir)importlib._bootstrap.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.importlib._bootstrap.c</DeepOutFile>
</None>
<None Include="..\Lib\importlib\_bootstrap_external.py">
<ModName>importlib._bootstrap_external</ModName>
<IntFile>$(IntDir)importlib._bootstrap_external.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\importlib._bootstrap_external.h</OutFile>
<DeepIntFile>$(IntDir)importlib._bootstrap_external.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.importlib._bootstrap_external.c</DeepOutFile>
</None>
<None Include="..\Lib\zipimport.py">
<ModName>zipimport</ModName>
<IntFile>$(IntDir)zipimport.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\zipimport.h</OutFile>
<DeepIntFile>$(IntDir)zipimport.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.zipimport.c</DeepOutFile>
</None>
<None Include="..\Lib\abc.py">
<ModName>abc</ModName>
<IntFile>$(IntDir)abc.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\abc.h</OutFile>
<DeepIntFile>$(IntDir)abc.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.abc.c</DeepOutFile>
</None>
<None Include="..\Lib\codecs.py">
<ModName>codecs</ModName>
<IntFile>$(IntDir)codecs.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\codecs.h</OutFile>
<DeepIntFile>$(IntDir)codecs.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.codecs.c</DeepOutFile>
</None>
<None Include="..\Lib\io.py">
<ModName>io</ModName>
<IntFile>$(IntDir)io.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\io.h</OutFile>
<DeepIntFile>$(IntDir)io.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.io.c</DeepOutFile>
</None>
<None Include="..\Lib\_collections_abc.py">
<ModName>_collections_abc</ModName>
<IntFile>$(IntDir)_collections_abc.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\_collections_abc.h</OutFile>
<DeepIntFile>$(IntDir)_collections_abc.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df._collections_abc.c</DeepOutFile>
</None>
<None Include="..\Lib\_sitebuiltins.py">
<ModName>_sitebuiltins</ModName>
<IntFile>$(IntDir)_sitebuiltins.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\_sitebuiltins.h</OutFile>
<DeepIntFile>$(IntDir)_sitebuiltins.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df._sitebuiltins.c</DeepOutFile>
</None>
<None Include="..\Lib\genericpath.py">
<ModName>genericpath</ModName>
<IntFile>$(IntDir)genericpath.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\genericpath.h</OutFile>
<DeepIntFile>$(IntDir)genericpath.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.genericpath.c</DeepOutFile>
</None>
<None Include="..\Lib\ntpath.py">
<ModName>ntpath</ModName>
<IntFile>$(IntDir)ntpath.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\ntpath.h</OutFile>
<DeepIntFile>$(IntDir)ntpath.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.ntpath.c</DeepOutFile>
</None>
<None Include="..\Lib\posixpath.py">
<ModName>posixpath</ModName>
<IntFile>$(IntDir)posixpath.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\posixpath.h</OutFile>
<DeepIntFile>$(IntDir)posixpath.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.posixpath.c</DeepOutFile>
</None>
<None Include="..\Lib\os.py">
<ModName>os</ModName>
<IntFile>$(IntDir)os.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\os.h</OutFile>
<DeepIntFile>$(IntDir)os.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.os.c</DeepOutFile>
</None>
<None Include="..\Lib\site.py">
<ModName>site</ModName>
<IntFile>$(IntDir)site.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\site.h</OutFile>
<DeepIntFile>$(IntDir)site.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.site.c</DeepOutFile>
</None>
<None Include="..\Lib\stat.py">
<ModName>stat</ModName>
<IntFile>$(IntDir)stat.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\stat.h</OutFile>
<DeepIntFile>$(IntDir)stat.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.stat.c</DeepOutFile>
</None>
<None Include="..\Lib\__hello__.py">
<ModName>__hello__</ModName>
<IntFile>$(IntDir)__hello__.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\__hello__.h</OutFile>
<DeepIntFile>$(IntDir)__hello__.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.__hello__.c</DeepOutFile>
</None>
<None Include="..\Lib\__phello__\__init__.py">
<ModName>__phello__</ModName>
<IntFile>$(IntDir)__phello__.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\__phello__.h</OutFile>
<DeepIntFile>$(IntDir)__phello__.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.__phello__.c</DeepOutFile>
</None>
<None Include="..\Lib\__phello__\ham\__init__.py">
<ModName>__phello__.ham</ModName>
<IntFile>$(IntDir)__phello__.ham.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\__phello__.ham.h</OutFile>
<DeepIntFile>$(IntDir)__phello__.ham.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.__phello__.ham.c</DeepOutFile>
</None>
<None Include="..\Lib\__phello__\ham\eggs.py">
<ModName>__phello__.ham.eggs</ModName>
<IntFile>$(IntDir)__phello__.ham.eggs.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\__phello__.ham.eggs.h</OutFile>
<DeepIntFile>$(IntDir)__phello__.ham.eggs.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.__phello__.ham.eggs.c</DeepOutFile>
</None>
<None Include="..\Lib\__phello__\spam.py">
<ModName>__phello__.spam</ModName>
<IntFile>$(IntDir)__phello__.spam.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\__phello__.spam.h</OutFile>
<DeepIntFile>$(IntDir)__phello__.spam.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.__phello__.spam.c</DeepOutFile>
</None>
<None Include="..\Tools\freeze\flag.py">
<ModName>frozen_only</ModName>
<IntFile>$(IntDir)frozen_only.g.h</IntFile>
<OutFile>$(PySourcePath)Python\frozen_modules\frozen_only.h</OutFile>
<DeepIntFile>$(IntDir)frozen_only.g.c</DeepIntFile>
<DeepOutFile>$(PySourcePath)Python\deepfreeze\df.frozen_only.c</DeepOutFile>
</None>
<!-- END frozen modules -->
</ItemGroup>
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
<ImportGroup Label="ExtensionTargets">
</ImportGroup>
<Target Name="_RebuildFrozen" AfterTargets="AfterBuild" Condition="$(Configuration) != 'PGUpdate'">
<Exec Command='"$(TargetPath)" "%(None.ModName)" "%(None.FullPath)" "%(None.IntFile)"' />
<Exec Command='"$(TargetPath)" "%(None.ModName)" "%(None.FullPath)" "%(None.DeepIntFile)"' />

<Copy SourceFiles="%(None.IntFile)"
<Copy SourceFiles="%(None.DeepIntFile)"
DestinationFiles="%(None.OutFile)"
Condition="!Exists(%(None.OutFile)) or (Exists(%(None.IntFile)) and '$([System.IO.File]::ReadAllText(%(None.OutFile)).Replace(`&#x0D;&#x0A;`, `&#x0A;`))' != '$([System.IO.File]::ReadAllText(%(None.IntFile)).Replace(`&#x0D;&#x0A;`, `&#x0A;`))')">
Condition="!Exists(%(None.OutFile)) or (Exists(%(None.DeepIntFile)) and '$([System.IO.File]::ReadAllText(%(None.OutFile)).Replace(`&#x0D;&#x0A;`, `&#x0A;`))' != '$([System.IO.File]::ReadAllText(%(None.DeepIntFile)).Replace(`&#x0D;&#x0A;`, `&#x0A;`))')">
<Output TaskParameter="CopiedFiles" ItemName="_Updated" />
</Copy>

<Message Text="Updated files: @(_Updated->'%(Filename)%(Extension)',', ')"
Condition="'@(_Updated)' != ''" Importance="high" />
</Target>
<Target Name="_RebuildDeepFrozen" AfterTargets="_RebuildFrozen" Condition="$(Configuration) != 'PGUpdate'">
<Exec Command='$(PythonForBuild) "$(PySourcePath)Tools\scripts\deepfreeze.py" "%(None.OutFile)" "-m" "%(None.ModName)" -o "%(None.IntFile)"' />
Copy link
Member Author

Choose a reason for hiding this comment

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

Do we need a separate DeepIntFile?

Copy link
Member

Choose a reason for hiding this comment

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

Probably, yeah. It'll be used for dependency analysis, and I think it's more likely to decide that it's always out of date than to incorrectly assume it's up to date, but it's safest to only have one task creating each file during a build.


<Copy SourceFiles="%(None.IntFile)"
DestinationFiles="%(None.DeepOutFile)"
Condition="!Exists(%(None.DeepOutFile)) or (Exists(%(None.IntFile)) and '$([System.IO.File]::ReadAllText(%(None.DeepOutFile)).Replace(`&#x0D;&#x0A;`, `&#x0A;`))' != '$([System.IO.File]::ReadAllText(%(None.IntFile)).Replace(`&#x0D;&#x0A;`, `&#x0A;`))')">
<Output TaskParameter="CopiedFiles" ItemName="_DeepUpdated" />
</Copy>

<Message Text="Updated files: @(_DeepUpdated->'%(Filename)%(Extension)',', ')"
Condition="'@(_DeepUpdated)' != ''" Importance="high" />
</Target>
<Target Name="_CleanFrozen" BeforeTargets="CoreClean" Condition="$(Configuration) != 'PGUpdate'">
<ItemGroup>
<Clean Include="%(None.IntFile)" />
Expand Down
6 changes: 3 additions & 3 deletions PCbuild/find_python.bat
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,13 +31,13 @@
@if "%_Py_EXTERNALS_DIR%"=="" (set _Py_EXTERNALS_DIR=%~dp0\..\externals)

@rem If we have Python in externals, use that one
@if exist "%_Py_EXTERNALS_DIR%\pythonx86\tools\python.exe" (set PYTHON="%_Py_EXTERNALS_DIR%\pythonx86\tools\python.exe") & (set _Py_Python_Source=found in externals directory) & goto :found
@if exist "%_Py_EXTERNALS_DIR%\pythonx86\tools\python.exe" ("%_Py_EXTERNALS_DIR%\pythonx86\tools\python.exe" -Ec "import sys; assert sys.version_info[:2] >= (3, 10)" >nul 2>nul) && (set PYTHON="%_Py_EXTERNALS_DIR%\pythonx86\tools\python.exe") && (set _Py_Python_Source=found in externals directory) && goto :found || rmdir /Q /S "%_Py_EXTERNALS_DIR%\pythonx86"

@rem If HOST_PYTHON is recent enough, use that
@if NOT "%HOST_PYTHON%"=="" @%HOST_PYTHON% -Ec "import sys; assert sys.version_info[:2] >= (3, 8)" >nul 2>nul && (set PYTHON="%HOST_PYTHON%") && (set _Py_Python_Source=found as HOST_PYTHON) && goto :found
@if NOT "%HOST_PYTHON%"=="" @%HOST_PYTHON% -Ec "import sys; assert sys.version_info[:2] >= (3, 10)" >nul 2>nul && (set PYTHON="%HOST_PYTHON%") && (set _Py_Python_Source=found as HOST_PYTHON) && goto :found

@rem If py.exe finds a recent enough version, use that one
@for %%p in (3.9 3.8) do @py -%%p -EV >nul 2>&1 && (set PYTHON=py -%%p) && (set _Py_Python_Source=found %%p with py.exe) && goto :found
@for %%p in (3.10) do @py -%%p -EV >nul 2>&1 && (set PYTHON=py -%%p) && (set _Py_Python_Source=found %%p with py.exe) && goto :found

@if NOT exist "%_Py_EXTERNALS_DIR%" mkdir "%_Py_EXTERNALS_DIR%"
@set _Py_NUGET=%NUGET%
Expand Down
24 changes: 24 additions & 0 deletions PCbuild/pythoncore.vcxproj
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -500,6 +500,30 @@
<ClCompile Include="..\Python\thread.c" />
<ClCompile Include="..\Python\traceback.c" />
</ItemGroup>
<ItemGroup>
<!-- BEGIN deepfreeze -->
<ClCompile Include="..\Python\deepfreeze\df.importlib._bootstrap.c" />
<ClCompile Include="..\Python\deepfreeze\df.importlib._bootstrap_external.c" />
<ClCompile Include="..\Python\deepfreeze\df.zipimport.c" />
<ClCompile Include="..\Python\deepfreeze\df.abc.c" />
<ClCompile Include="..\Python\deepfreeze\df.codecs.c" />
<ClCompile Include="..\Python\deepfreeze\df.io.c" />
<ClCompile Include="..\Python\deepfreeze\df._collections_abc.c" />
<ClCompile Include="..\Python\deepfreeze\df._sitebuiltins.c" />
<ClCompile Include="..\Python\deepfreeze\df.genericpath.c" />
<ClCompile Include="..\Python\deepfreeze\df.ntpath.c" />
<ClCompile Include="..\Python\deepfreeze\df.posixpath.c" />
<ClCompile Include="..\Python\deepfreeze\df.os.c" />
<ClCompile Include="..\Python\deepfreeze\df.site.c" />
<ClCompile Include="..\Python\deepfreeze\df.stat.c" />
<ClCompile Include="..\Python\deepfreeze\df.__hello__.c" />
<ClCompile Include="..\Python\deepfreeze\df.__phello__.c" />
<ClCompile Include="..\Python\deepfreeze\df.__phello__.ham.c" />
<ClCompile Include="..\Python\deepfreeze\df.__phello__.ham.eggs.c" />
<ClCompile Include="..\Python\deepfreeze\df.__phello__.spam.c" />
<ClCompile Include="..\Python\deepfreeze\df.frozen_only.c" />
<!-- END deepfreeze -->
</ItemGroup>
<ItemGroup Condition="$(IncludeExternals)">
<ClCompile Include="..\Modules\zlibmodule.c" />
<ClCompile Include="$(zlibDir)\adler32.c" />
Expand Down
5 changes: 0 additions & 5 deletions Python/frozen.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -61,12 +61,7 @@
#include "frozen_modules/frozen_only.h"
/* End includes */

#ifdef MS_WINDOWS
/* Deepfreeze isn't supported on Windows yet. */
#define GET_CODE(name) NULL
#else
#define GET_CODE(name) _Py_get_##name##_toplevel
#endif

/* Start extern declarations */
extern PyObject *_Py_get_importlib__bootstrap_toplevel(void);
Expand Down
58 changes: 46 additions & 12 deletions Tools/scripts/deepfreeze.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,13 +1,16 @@
import argparse
import ast
import builtins
import collections
import contextlib
import os
import sys
import re
import time
import types
import typing

import umarshal

verbose = False


Expand Down Expand Up @@ -55,7 +58,8 @@ def get_localsplus_counts(code: types.CodeType,
nplaincellvars += 1
elif kind & CO_FAST_FREE:
nfreevars += 1
assert nlocals == len(code.co_varnames) == code.co_nlocals
assert nlocals == len(code.co_varnames) == code.co_nlocals, \
(nlocals, len(code.co_varnames), code.co_nlocals)
assert ncellvars == len(code.co_cellvars)
assert nfreevars == len(code.co_freevars)
assert len(names) == nlocals + nplaincellvars + nfreevars
Expand Down Expand Up @@ -274,14 +278,7 @@ def generate_tuple(self, name: str, t: tuple[object, ...]) -> str:
self.write(item + ",")
return f"& {name}._object.ob_base.ob_base"

def generate_int(self, name: str, i: int) -> str:
maxint = sys.maxsize
if maxint == 2**31 - 1:
digit = 2**15
elif maxint == 2**63 - 1:
digit = 2**30
else:
assert False, f"What int size is this system?!? {maxint=}"
def _generate_int_for_bits(self, name: str, i: int, digit: int) -> None:
sign = -1 if i < 0 else 0 if i == 0 else +1
i = abs(i)
digits: list[int] = []
Expand All @@ -298,6 +295,20 @@ def generate_int(self, name: str, i: int) -> str:
if digits:
ds = ", ".join(map(str, digits))
self.write(f".ob_digit = {{ {ds} }},")

def generate_int(self, name: str, i: int) -> str:
if abs(i) < 2**15:
self._generate_int_for_bits(name, i, 2**15)
else:
connective = "if"
for bits_in_digit in 15, 30:
self.write(f"#{connective} PYLONG_BITS_IN_DIGIT == {bits_in_digit}")
self._generate_int_for_bits(name, i, 2**bits_in_digit)
connective = "elif"
self.write("#else")
self.write('#error "PYLONG_BITS_IN_DIGIT should be 15 or 30"')
self.write("#endif")
# If neither clause applies, it won't compile
return f"& {name}.ob_base.ob_base"

def generate_float(self, name: str, x: float) -> str:
Expand Down Expand Up @@ -326,7 +337,7 @@ def generate(self, name: str, obj: object) -> str:
return self.cache[key]
self.misses += 1
match obj:
case types.CodeType() as code:
case types.CodeType() | umarshal.Code() as code:
val = self.generate_code(name, code)
case tuple(t):
val = self.generate_tuple(name, t)
Expand Down Expand Up @@ -367,8 +378,31 @@ def generate(self, name: str, obj: object) -> str:
}
"""

FROZEN_COMMENT = "/* Auto-generated by Programs/_freeze_module.c */"

FROZEN_DATA_LINE = r"\s*(\d+,\s*)+\s*"


def is_frozen_header(source: str) -> bool:
return source.startswith(FROZEN_COMMENT)


def decode_frozen_data(source: str) -> types.CodeType:
lines = source.splitlines()
while lines and re.match(FROZEN_DATA_LINE, lines[0]) is None:
del lines[0]
while lines and re.match(FROZEN_DATA_LINE, lines[-1]) is None:
del lines[-1]
values: tuple[int, ...] = ast.literal_eval("".join(lines))
data = bytes(values)
return umarshal.loads(data)


def generate(source: str, filename: str, modname: str, file: typing.TextIO) -> None:
code = compile(source, filename, "exec")
if is_frozen_header(source):
code = decode_frozen_data(source)
else:
code = compile(source, filename, "exec")
printer = Printer(file)
printer.generate("toplevel", code)
printer.write("")
Expand Down
Loading