gh-128627: Use __builtin_wasm_test_function_pointer_signature for Emscripten trampoline by hoodmane · Pull Request #137470 · python/cpython · GitHub

Since llvm/llvm-project#150201 was merged, there is now a better way to do this. Requires Emscripten 4.0.12. This lets us get rid of all the hand encoded webassembly.

@freakboy3742 this is blocked on updating the buildbot to use Emscripten 4.0.12, not sure what that entails.

Issue: Emscripten: Use wasm-gc for call trampolines instead of JS type reflection #128627

…or Emscripten trampolines Since llvm/llvm-project#150201 was merged, there is now a better way to do this. Requires Emscripten 4.0.12.

!buildbot emscripten

🤖 New build scheduled with the buildbot fleet by @hoodmane for commit 0ffb76b 🤖

Results will be shown at:

https://buildbot.python.org/all/#/grid?branch=refs%2Fpull%2F137470%2Fmerge

The command will test the builders whose names match following regular expression: emscripten

The builders matched are:

WASM Emscripten PR

!buildbot emscripten

🤖 New build scheduled with the buildbot fleet by @hoodmane for commit 04e0541 🤖

Results will be shown at:

https://buildbot.python.org/all/#/grid?branch=refs%2Fpull%2F137470%2Fmerge

The command will test the builders whose names match following regular expression: emscripten

The builders matched are:

WASM Emscripten PR

There's three blockers to this at present:

The check-c-globals analyser check that is failing CI on this PR
The test suite CI failure that is also visible on main, because of the change from gh-131876: extract _hashlib helpers into a separate directory #136995. gh-131876: extract _hashlib helpers into a separate directory #137319 looks like a possible fix; however, it looks like it's on hold pending some other changes.
The refactoring of the build script that we discussed at EuroPython (and on the buildmaster-config PR) to encode the Emscripten version as part of the build script, and use an argument to point at a local cache directory.

The alternative to (3) would be to violate the "RC1 is where we lock in the Emscripten version" rule, and bump to 4.0.12. That isn't an approach we could take in the general case, but given this is the first "back to Tier 3" release, and there's no impact on downstream packagers (since nobody can publish to PyPI, so nobody will be relying on the 4.0.11 ABI guarantee), I could probably be convinced to do this - but I'd like @hugovk to sign off on the plan as release manager.

Of course (3) is going to be needed in the fullness of time - it's just a question of whether we need to do it right now. And (1) and (2) will need a fix regardless.

Marking as "needs changes" - see previous comment.

A Python core developer has requested some changes be made to your pull request before we can consider merging it. If you could please address their requests along with any other requests in other reviews from core developers that would be appreciated.

Once you have made the requested changes, please leave a comment on this pull request containing the phrase I have made the requested changes; please review again. I will then notify any core developers who have left a review that you're ready for them to take another look at this pull request.

This PR isn't urgent so I think it would be fine to block it on making an Emscripten version manager. But it would also be fine to bump the Emscripten version for 3.14 imo.

I reverted the hashlib refactorization by the way, so failures on main shouldn't be because of this.

Thanks @picnixz. Failure on main is a runtime failure in a test about calendars and locales, not the linker error.

Hmm the C parser doesn't work with out of tree builds because it passes a bunch of invalid paths like -I/home/rchatham/cpython/*./Include cc @ericsnowcurrently
https://github.com/python/cpython/blob/main/Tools/c-analyzer/cpython/_parser.py?plain=1#L120

I can't reproduce the NotImplementedError locally:

NotImplementedError: (Struct(file=FileInfo(filename='/home/runner/work/cpython/cpython/Include/pystate.h', lno=62), name='_stack_chunk', data=[Member(name='previous', vartype=VarType(typequal=None, typespec='struct _stack_chunk', abstract='*'), size=None), Member(name='size', vartype=VarType(typequal=None, typespec='size_t', abstract=''), size=None), Member(name='top', vartype=VarType(typequal=None, typespec='size_t', abstract=''), size=None), Member(name='data', vartype=VarType(typequal=None, typespec='PyObject', abstract='* [1]'), size=None)], parent=None), [TypeDef(file=FileInfo(filename='/home/runner/work/cpython/cpython/Include/pytypedefs.h', lno=18), name='PyObject', data=VarType(typequal=None, typespec='struct _object', abstract=''), parent=None), TypeDef(file=FileInfo(filename='/home/runner/work/cpython/cpython/Python/emscripten_trampoline_inner.c', lno=1), name='PyObject', data=VarType(typequal=None, typespec='void', abstract=''), parent=None)])

I reverted the hashlib refactorization by the way, so failures on main shouldn't be because of this.

My apologies - I missed that the cause of the CI failures on main had changed. Thanks for the revert - now to work out what's going on with Calendars... :-)

With the exception of the generated files failure, this looks good to me.

Would be worth adding a short comment about why this is needed.

!buildbot emscripten

🤖 New build scheduled with the buildbot fleet by @freakboy3742 for commit a1753d6 🤖

Results will be shown at:

https://buildbot.python.org/all/#/grid?branch=refs%2Fpull%2F137470%2Fmerge

The command will test the builders whose names match following regular expression: emscripten

The builders matched are:

WASM Emscripten PR

!buildbot emscripten

🤖 New build scheduled with the buildbot fleet by @hoodmane for commit 0ca9e4c 🤖

Results will be shown at:

https://buildbot.python.org/all/#/grid?branch=refs%2Fpull%2F137470%2Fmerge

The command will test the builders whose names match following regular expression: emscripten

The builders matched are:

WASM Emscripten PR

LGTM!

Thanks @hoodmane for the PR, and @freakboy3742 for merging it 🌮🎉.. I'm working now to backport this PR to: 3.14.
🐍🍒⛏🤖

…or Emscripten trampoline (pythonGH-137470) With llvm/llvm-project#150201 being merged, there is now a better way to generate the Emscripten trampoline, instead of including hand-generated binary WASM content. Requires Emscripten 4.0.12. (cherry picked from commit 2629ee4) Co-authored-by: Hood Chatham <roberthoodchatham@gmail.com>

GH-139039 is a backport of this pull request to the 3.14 branch.

…for Emscripten trampoline (GH-137470) (#139039) gh-128627: Use __builtin_wasm_test_function_pointer_signature for Emscripten trampoline (GH-137470) With llvm/llvm-project#150201 being merged, there is now a better way to generate the Emscripten trampoline, instead of including hand-generated binary WASM content. Requires Emscripten 4.0.12. (cherry picked from commit 2629ee4) Co-authored-by: Hood Chatham <roberthoodchatham@gmail.com>

hoodmane requested review from corona10 and erlend-aasland as code owners August 6, 2025 15:35

bedevere-app Bot added the awaiting review label Aug 6, 2025

bedevere-app Bot mentioned this pull request Aug 6, 2025

Emscripten: Use wasm-gc for call trampolines instead of JS type reflection #128627

Closed

hoodmane added the skip news label Aug 6, 2025

pythongh-128627 : Use __builtin_wasm_test_function_pointer_signature f…

0ffb76b

…or Emscripten trampolines Since llvm/llvm-project#150201 was merged, there is now a better way to do this. Requires Emscripten 4.0.12.

hoodmane force-pushed the __builtin_wasm_test_function_pointer_signature branch from 6bbee1e to 0ffb76b Compare August 6, 2025 15:41

hoodmane requested a review from freakboy3742 August 6, 2025 15:42

Add forgotten file

04e0541

freakboy3742 reviewed Aug 6, 2025

View reviewed changes

Comment thread Makefile.pre.in

freakboy3742 requested changes Aug 6, 2025

View reviewed changes

bedevere-app Bot removed the awaiting review label Aug 6, 2025

bedevere-app Bot added the awaiting changes label Aug 6, 2025

hoodmane added 2 commits August 7, 2025 10:38

Mark global as const

b633734

Add prefix to globals

538e563

hoodmane requested a review from AA-Turner as a code owner August 7, 2025 08:41

hoodmane added 2 commits August 7, 2025 10:46

Add a comment

7ed9118

Update comment

fff815c

hoodmane requested a review from ZeroIntensity as a code owner September 17, 2025 08:40

hoodmane added 5 commits September 17, 2025 10:52

More comments

3379e1c

Cleanup

7d5a013

Reduce diff

b80911b

Reduce diff

b4f07d9

Cleanup

51af9e1

freakboy3742 requested changes Sep 17, 2025

View reviewed changes

Comment thread Makefile.pre.in Outdated

typedef => #define

a1753d6

freakboy3742 reviewed Sep 17, 2025

View reviewed changes

Fix again

0ca9e4c

Add comment

4002452

freakboy3742 approved these changes Sep 17, 2025

View reviewed changes

bedevere-app Bot added awaiting merge and removed awaiting changes labels Sep 17, 2025

freakboy3742 merged commit 2629ee4 into python:main Sep 17, 2025
46 of 47 checks passed

bedevere-app Bot removed the awaiting merge label Sep 17, 2025

freakboy3742 added awaiting merge needs backport to 3.14 bugs and security fixes labels Sep 17, 2025

bedevere-app Bot removed the needs backport to 3.14 bugs and security fixes label Sep 17, 2025

This was referenced Feb 20, 2026

Emscripten: Trampoline doesn't work with emcc 4.0.19 due to table/memory visibility #145037

Closed

gh-145037: Fix Emscripten trampoline with emcc >= 4.0.19 #145038

Merged

Fix wasm-gc call trampoline when RELOCATABLE isn't passed pyodide/pyodide#6115

Merged

+              // support wasm-gc yet. If the JS runtime does not support wasm-gc (or has buggy
+              // support like iOS), we will use the JS trampoline fallback.
+              #define PyObject void;

gh-128627: Use __builtin_wasm_test_function_pointer_signature for Emscripten trampoline by hoodmane · Pull Request #137470 · python/cpython · GitHub

Uh oh!

Uh oh!

gh-128627: Use __builtin_wasm_test_function_pointer_signature for Emscripten trampoline - #137470

Conversation

hoodmane commented Aug 6, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

hoodmane commented Aug 6, 2025

Uh oh!

bedevere-bot commented Aug 6, 2025

Uh oh!

hoodmane commented Aug 6, 2025

Uh oh!

bedevere-bot commented Aug 6, 2025

Uh oh!

Uh oh!

freakboy3742 commented Aug 6, 2025

Uh oh!

freakboy3742 left a comment

Choose a reason for hiding this comment

Uh oh!

bedevere-app Bot commented Aug 6, 2025

Uh oh!

hoodmane commented Aug 7, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

picnixz commented Aug 7, 2025

Uh oh!

hoodmane commented Aug 7, 2025

Uh oh!

hoodmane commented Aug 7, 2025

Uh oh!

hoodmane commented Aug 7, 2025

Uh oh!

freakboy3742 commented Aug 8, 2025

Uh oh!

freakboy3742 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

freakboy3742 Sep 17, 2025

Choose a reason for hiding this comment

Uh oh!

freakboy3742 commented Sep 17, 2025

Uh oh!

bedevere-bot commented Sep 17, 2025

Uh oh!

hoodmane commented Sep 17, 2025

Uh oh!

bedevere-bot commented Sep 17, 2025

Uh oh!

freakboy3742 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

miss-islington-app Bot commented Sep 17, 2025

Uh oh!

bedevere-app Bot commented Sep 17, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

hoodmane commented Aug 6, 2025 •
edited
Loading

hoodmane commented Aug 7, 2025 •
edited
Loading