nickdesaulniers wrote: The front end should boil "g" down to "mir" (in some order). Maybe separate checks along the lines "string contains 'r'" and "string contains 'm'". I swear I added a flag to InlineAsm (is spillable) or something, but can't recall if that's for the MIR or middle end. Maybe whatever I'm thinking of can be reused here?
On Wed, Feb 4, 2026, 11:52 AM Bill Wendling ***@***.***> wrote: > ***@***.**** commented on this pull request. > ------------------------------ > > In clang/lib/CodeGen/CGStmt.cpp > <https://github.com/llvm/llvm-project/pull/92040#discussion_r2765562342>: > > > QualType QTy = OutExpr->getType(); > const bool IsScalarOrAggregate = hasScalarEvaluationKind(QTy) || > hasAggregateEvaluationKind(QTy); > - if (!Info.allowsMemory() && IsScalarOrAggregate) { > + const bool RegisterMemoryConstraints = > + OutputConstraint == "rm" || OutputConstraint == "mr"; > > It's not a bad idea, and most likely the correct way to handle this. But > I'm afraid of picking up constraints that aren't "rm", like "g" (which we > should handle...). I can add a FIXME there to expand it in the future? > > — > Reply to this email directly, view it on GitHub > <https://github.com/llvm/llvm-project/pull/92040#discussion_r2765562342>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AAN5IXZH44MRF42I5AF2HZT4KJEYNAVCNFSM6AAAAACPUWSYHGVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZTONJSHA2DSOJXHE> > . > You are receiving this because your review was requested.Message ID: > ***@***.***> > https://github.com/llvm/llvm-project/pull/92040 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
