Normalize out-of-range integer literal casts - #180
Open
nikswamy wants to merge 1 commit into
Open
Conversation
Generated C interfaces commonly define fixed-width bit-pattern constants as explicit casts from long integer literals and expand them into switch tables. On LP64 hosts, a high-bit long literal is a signed 64-bit value, while casting it to int32_t denotes a negative target bit pattern under PAL's two's-complement model. Emitting the source literal followed by Int.Cast.int64_to_int32 leaves F* with an out-of-range conversion and prevents otherwise routine generated switches from verifying. Clang exposes integer literal bit patterns through a signed BigInt representation, even when the C literal type is unsigned. Recover the source literal mathematical value from its signedness and width before deciding whether a cast is representable in the target type. Preserve representable casts as explicit F* Int.Cast operations: these conversions have defined C semantics and retaining the conversion term is useful to downstream proofs. For out-of-range casts, normalize to PAL's target representation and emit a target-typed literal directly. Use one machine-literal emitter for ordinary literals and folded casts. Emit native F* suffixes for every supported C width: y/uy, s/us, l/ul, and L/UL, with parenthesized negative signed literals and constructor fallbacks for non-standard widths. Add a dedicated regression covering direct and X-macro-generated switch tables of high-bit signed 32-bit codes, all eight suffix forms, narrowing and wraparound at every width, UINT64_MAX converted to signed int64, and a representable int64-to-uint64 cast that remains explicit. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 981f7d3c-6a91-47ad-84d7-7aadf747123d
gebner
reviewed
Jul 23, 2026
| if !integer_fits(&literal_value, *signed, *width) { | ||
| return emit_machine_int_literal(&literal_value, *signed, *width); | ||
| } | ||
| } |
Contributor
There was a problem hiding this comment.
This should probably go in a separate cast normalization pass probably,
gebner
reviewed
Jul 23, 2026
| } => normalize_unsigned(val, width), | ||
| _ => val.as_ref().clone(), | ||
| }; | ||
| if !integer_fits(&literal_value, *signed, *width) { |
Contributor
There was a problem hiding this comment.
It's a bit surprising to gate this on fitting: this simplification is always valid, right?
Contributor
There was a problem hiding this comment.
After reading up on this, the simplification is in general only valid if it fits (but not in the special case implemented by this PR).
Unsigned-to-signed casts are implementation-defined when the value doesn't fit. I'm happy to accept the restriction to two's-complement though.
gebner
requested changes
Jul 23, 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
2 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.
Generated C interfaces commonly define fixed-width bit-pattern constants as explicit casts from long integer literals and expand them into switch tables. On LP64 hosts, a high-bit long literal is a signed 64-bit value, while casting it to int32_t denotes a negative target bit pattern under PAL's two's-complement model. Emitting the source literal followed by Int.Cast.int64_to_int32 leaves F* with an out-of-range conversion and prevents otherwise routine generated switches from verifying.
Clang exposes integer literal bit patterns through a signed BigInt representation, even when the C literal type is unsigned. Recover the source literal mathematical value from its signedness and width before deciding whether a cast is representable in the target type.
Preserve representable casts as explicit F* Int.Cast operations: these conversions have defined C semantics and retaining the conversion term is useful to downstream proofs. For out-of-range casts, normalize to PAL's target representation and emit a target-typed literal directly.
Use one machine-literal emitter for ordinary literals and folded casts. Emit native F* suffixes for every supported C width: y/uy, s/us, l/ul, and L/UL, with parenthesized negative signed literals and constructor fallbacks for non-standard widths.
Add a dedicated regression covering direct and X-macro-generated switch tables of high-bit signed 32-bit codes, all eight suffix forms, narrowing and wraparound at every width, UINT64_MAX converted to signed int64, and a representable int64-to-uint64 cast that remains explicit.