[PATCH] D59373: Make a hack for LTO work only when you are actually doing LTO.

[Eli Friedman via Phabricator via llvm-commits]

efriedma accepted this revision.
efriedma added a comment.
This revision is now accepted and ready to land.

LGTM



================
Comment at: lld/ELF/Driver.cpp:220
+        if (identify_magic(P.first.getBuffer()) != file_magic::bitcode)
+          return;
+
----------------
ruiu wrote:
> efriedma wrote:
> > "return"?  It seems weird to intentionally ignore archive files without a symbol table.
> But that is what other linkers do and is also a natural consequence of how linker works on archive files. When a linker visits an archive file, it reads its symbol table, and pulls out object files that defines symbols that are currently undefined. Therefore, if an archive doesn't contain a symbol table, nothing is pulled out, and the archive is effectively ignored.
I guess that's right; it's effectively the same as the other codepath which calls `make<ArchiveFile>()`.  But it looks weird at first glance; maybe add an explicit comment noting that.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D59373/new/

https://reviews.llvm.org/D59373