efriedma added a comment.
> Would it be reasonable to instead start with a stopgap measure of not adding
> attributes for this/return of thunks?
You mean, add align attributes to "this" on regular methods, but not thunks? I
can't see how that could do any harm.
================
Comment at: clang/lib/CodeGen/CGVTables.cpp:467
+ if (!CalleeMD->isDefined()) {
+ CGM.ErrorUnsupported(ThunkGD.getDecl(), "thunk for forward declaration");
+ return;
----------------
"IsUnprototyped" means that we have to generate a thunk, but we don't have the
proper types. We emit some special IR with known incorrect types, and mark it
with the "thunk" attribute to let the optimizer know what we're doing.
It should be fine to continue using a fake type on the IsUnprototyped codepath.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D100388/new/
https://reviews.llvm.org/D100388
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
