fix[codegen]: widen ternary arms to the result type layout - #5236
Merged
Merged
Conversation
Gas ChangesNo changes detected. Summary
|
📊 Bytecode Size Changes (venom)No changes detected. Full bytecode sizes
|
harkal
changed the title
harkal
marked this pull request as ready for review
September 3, 2026 11:12
Sporarum
requested changes
Sep 3, 2026
Sporarum
left a comment
Sporarum
left a comment
Collaborator
There was a problem hiding this comment.
Good find !
It might be possible to remove the need for a lot of added logic by setting the type in metadata in a smarter way instead, see one of the comments below
Comment on lines
+1227
to
+1251
| def punnable(src_typ, dst_typ): | ||
| """ | ||
| Return True if memory laid out as `src_typ` can be read (punned) as `dst_typ`. | ||
|
|
||
| Compatible types can differ in element or member layout, e.g. | ||
| DynArray[Bytes[10], 5] vs DynArray[Bytes[512], 5] (element stride 64 | ||
| vs 544). A wider bytestring bound or DynArray capacity at the top level | ||
| does not change the layout of the data that is present. | ||
| """ | ||
| if isinstance(dst_typ, (DArrayT, SArrayT)): | ||
| assert isinstance(src_typ, (DArrayT, SArrayT)) | ||
| pairs = [(src_typ.value_type, dst_typ.value_type)] | ||
| elif isinstance(dst_typ, TupleT): | ||
| assert isinstance(src_typ, TupleT) | ||
| n = len(dst_typ.member_types) | ||
| assert len(src_typ.member_types) == n | ||
| pairs = [(src_typ.member_types[i], dst_typ.member_types[i]) for i in range(n)] | ||
| else: | ||
| # primitive words, bytestrings ([length][data]) and (nominal) structs | ||
| return True | ||
|
|
||
| # nested values also need equal sizes: they determine strides and offsets | ||
| return all( | ||
| s.memory_bytes_required == d.memory_bytes_required and punnable(s, d) for s, d in pairs | ||
| ) |
Collaborator
There was a problem hiding this comment.
I think this can be simplified by moving the length check to the beginning:
Suggested change
| def punnable(src_typ, dst_typ): | |
| """ | |
| Return True if memory laid out as `src_typ` can be read (punned) as `dst_typ`. | |
| Compatible types can differ in element or member layout, e.g. | |
| DynArray[Bytes[10], 5] vs DynArray[Bytes[512], 5] (element stride 64 | |
| vs 544). A wider bytestring bound or DynArray capacity at the top level | |
| does not change the layout of the data that is present. | |
| """ | |
| if isinstance(dst_typ, (DArrayT, SArrayT)): | |
| assert isinstance(src_typ, (DArrayT, SArrayT)) | |
| pairs = [(src_typ.value_type, dst_typ.value_type)] | |
| elif isinstance(dst_typ, TupleT): | |
| assert isinstance(src_typ, TupleT) | |
| n = len(dst_typ.member_types) | |
| assert len(src_typ.member_types) == n | |
| pairs = [(src_typ.member_types[i], dst_typ.member_types[i]) for i in range(n)] | |
| else: | |
| # primitive words, bytestrings ([length][data]) and (nominal) structs | |
| return True | |
| # nested values also need equal sizes: they determine strides and offsets | |
| return all( | |
| s.memory_bytes_required == d.memory_bytes_required and punnable(s, d) for s, d in pairs | |
| ) | |
| def punnable(src_typ, dst_typ): | |
| """ | |
| Return True if memory laid out as `src_typ` can be read (punned) as `dst_typ`. | |
| Compatible types can differ in element or member layout, e.g. | |
| DynArray[Bytes[10], 5] vs DynArray[Bytes[512], 5] (element stride 64 | |
| vs 544). A wider bytestring bound or DynArray capacity at the top level | |
| does not change the layout of the data that is present. | |
| """ | |
| if src_typ.memory_bytes_required == dst_typ.memory_bytes_required: | |
| return False | |
| if isinstance(dst_typ, (DArrayT, SArrayT)): | |
| assert isinstance(src_typ, (DArrayT, SArrayT)) | |
| pairs = [(src_typ.value_type, dst_typ.value_type)] | |
| elif isinstance(dst_typ, TupleT): | |
| assert isinstance(src_typ, TupleT) | |
| n = len(dst_typ.member_types) | |
| assert len(src_typ.member_types) == n | |
| pairs = [(src_typ.member_types[i], dst_typ.member_types[i]) for i in range(n)] | |
| else: | |
| # primitive words, bytestrings ([length][data]) and (nominal) structs | |
| return True | |
| # nested values also need equal sizes: they determine strides and offsets | |
| return all(punnable(s, d) for s, d in pairs) |
Which then allows us to recurse immediately:
Suggested change
| def punnable(src_typ, dst_typ): | |
| """ | |
| Return True if memory laid out as `src_typ` can be read (punned) as `dst_typ`. | |
| Compatible types can differ in element or member layout, e.g. | |
| DynArray[Bytes[10], 5] vs DynArray[Bytes[512], 5] (element stride 64 | |
| vs 544). A wider bytestring bound or DynArray capacity at the top level | |
| does not change the layout of the data that is present. | |
| """ | |
| if isinstance(dst_typ, (DArrayT, SArrayT)): | |
| assert isinstance(src_typ, (DArrayT, SArrayT)) | |
| pairs = [(src_typ.value_type, dst_typ.value_type)] | |
| elif isinstance(dst_typ, TupleT): | |
| assert isinstance(src_typ, TupleT) | |
| n = len(dst_typ.member_types) | |
| assert len(src_typ.member_types) == n | |
| pairs = [(src_typ.member_types[i], dst_typ.member_types[i]) for i in range(n)] | |
| else: | |
| # primitive words, bytestrings ([length][data]) and (nominal) structs | |
| return True | |
| # nested values also need equal sizes: they determine strides and offsets | |
| return all( | |
| s.memory_bytes_required == d.memory_bytes_required and punnable(s, d) for s, d in pairs | |
| ) | |
| def punnable(src_typ, dst_typ): | |
| """ | |
| Return True if memory laid out as `src_typ` can be read (punned) as `dst_typ`. | |
| Compatible types can differ in element or member layout, e.g. | |
| DynArray[Bytes[10], 5] vs DynArray[Bytes[512], 5] (element stride 64 | |
| vs 544). A wider bytestring bound or DynArray capacity at the top level | |
| does not change the layout of the data that is present. | |
| """ | |
| if src_typ.memory_bytes_required == dst_typ.memory_bytes_required: | |
| return False | |
| if isinstance(dst_typ, (DArrayT, SArrayT)): | |
| assert isinstance(src_typ, (DArrayT, SArrayT)) | |
| return punnable(src_typ.value_type, dst_typ.value_type) | |
| elif isinstance(dst_typ, TupleT): | |
| assert isinstance(src_typ, TupleT) | |
| assert len(src_typ.member_types) == len(dst_typ.member_types) | |
| return all(punnable(s, d) for s,d in zip(src_typ.member_types, dst_typ.member_types)) | |
| else: | |
| # primitive words, bytestrings ([length][data]) and (nominal) structs | |
| return True |
Collaborator
Author
There was a problem hiding this comment.
The top-level size is deliberately not compared. Bytes[10] vs Bytes[40] and DynArray[uint256, 2] vs DynArray[uint256, 5] are punnable despite different sizes (the same-type test asserts no copy for those), and size_in_bytes panics for a DynArray[..., INF] result. Sizes only matter one level down, where they fix the stride.
Also the suggested early return is inverted (returns False on equal sizes), and the second version brings zip back, which @charles-cooper asked to remove above.
HodanPlodky
approved these changes
Sep 3, 2026
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 22ada0e45d
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
Sporarum
reviewed
Sep 4, 2026
Sporarum
approved these changes
Sep 4, 2026
harkal
enabled auto-merge (squash)
September 4, 2026 07:18
harkal
disabled auto-merge
September 4, 2026 07:18
harkal
enabled auto-merge (squash)
September 4, 2026 07:18
This was referenced Sep 4, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What I did
Fix a silent miscompile in both codegen pipelines: a ternary (
x if c else y) over an aggregate returned data laid out per the arm's type but read with the widened result type. For examplereturned
[b"a", b"", b""]fora = [b"a", b"0123456789", b""]. Same for nested DynArray capacity widening, static arrays of DynArrays, String elements and tuple members (via locals or internal call results). Legacy and venom codegen are both affected.How I did it
Each arm is converted to the result layout inside its own branch, before the join. Legacy
parse_IfExpcopies the arm into an internal variable of the result type withmake_setter; venomlower_IfExpuses its typed memory store. Both already widen element-wise.The gate is a shared structural layout check,
same_memory_layoutincodegen/core.py, not type equality. Same-type arms, bytestring arms with different bounds, capacity-only widening and unbounded result types produce byte-identical IR to master on both pipelines.How to verify it
Commit message
Description for the changelog
Fix ternary expressions over aggregates whose arms are narrower than the result type.
Cute Animal Picture