[Sema][SYCL] Only defer device diags for constexpr var init under SYCL - #22767
Open
wenju-he wants to merge 2 commits into
Open
[Sema][SYCL] Only defer device diags for constexpr var init under SYCL#22767wenju-he wants to merge 2 commits into
wenju-he wants to merge 2 commits into
Conversation
InConstexprVarInit gates a mechanism (added for SYCL's "manifestly constant-evaluated" device-code exemption) that reroutes deferred device diagnostics into MaybeDeviceDeferredDiags, discarding them if the variable turns out to be constant-initialized. It was originally opt-in via LangOpts.SYCLAllowAllFeaturesInConstexpr (added in fa62920), but 10fee97 ("[SYCL][clang] Always lift SYCL device code restrictions in constant expressions") made it unconditional for any constexpr-capable global/static variable initializer, so it also swallowed CUDA/HIP deferred diagnostics (e.g. AMDGPU builtin argument checks) raised while initializing a global constexpr lambda, since SemaBase::SemaDiagnosticBuilder routes to whichever map InConstexprVarInit currently selects regardless of which DiagIfDeviceCode caller (SYCL or CUDA) created the deferred diagnostic. Restrict it to SYCL device compilation, its only real consumer (SemaSYCL::DiagIfDeviceCode). Fix clang/test/SemaHIP/amdgpu-builtin-in-lambda.hip CMPLRLLVM-77209 Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
elizabethandrews
requested changes
Jul 28, 2026
elizabethandrews
left a comment
elizabethandrews
left a comment
Contributor
There was a problem hiding this comment.
I do not think we should gate where InConstexprVarInit is set. It accurately records that are parsing a constexpr variable's initializer. We should gate it's consumption.
…rInit Per review, InConstexprVarInit should stay an accurate, ungated record of "we are parsing a constexpr variable's initializer" rather than being restricted to SYCL device compilation. Instead, gate its consumption: SemaBase::SemaDiagnosticBuilder::getDeviceDeferredDiags() now only reroutes to MaybeDeviceDeferredDiags when SYCLIsDevice is also set, so CUDA/HIP deferred diagnostics (e.g. AMDGPU builtin argument checks) are no longer swallowed by SYCL's "manifestly constant-evaluated" exemption. CMPLRLLVM-77209 Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Contributor
Author
done |
tahonermann
approved these changes
Jul 29, 2026
Contributor
There was a problem hiding this comment.
Looks good with the update to address @elizabethandrews's comment. Thanks @wenju-he!
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
3 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.
InConstexprVarInit marks that Sema is parsing a possibly
constant-evaluated variable initializer and should stay accurate for
all languages. What's SYCL-specific is SemaSYCL's "manifestly
constant-evaluated" exemption, which reroutes device diagnostics into
MaybeDeviceDeferredDiags and discards them if the variable turns out
constant-initialized. 10fee97 made InConstexprVarInit
unconditional, so this also started swallowing CUDA/HIP deferred
diagnostics (e.g. AMDGPU builtin checks in a global constexpr lambda).
Gate the consumption instead: getDeviceDeferredDiags() only reroutes
to MaybeDeviceDeferredDiags when SYCLIsDevice is set.
Fix clang/test/SemaHIP/amdgpu-builtin-in-lambda.hip
CMPLRLLVM-77209