[PATCH] D79900: [BitcodeReader] datalayout must be specified before it is queried.

[Eli Friedman via Phabricator via llvm-commits]

efriedma marked an inline comment as done.
efriedma added inline comments.


================
Comment at: llvm/lib/Bitcode/Reader/BitcodeReader.cpp:3445
+    TheModule->setDataLayout(DL);
+  };
+
----------------
jdoerfert wrote:
> mehdi_amini wrote:
> > jdoerfert wrote:
> > > efriedma wrote:
> > > > jdoerfert wrote:
> > > > > Should we emit an error here? I mean, can it happen we now silently ignore the DL or triple and the user is baffled what is happening?
> > > > We print an error if we see MODULE_CODE_DATALAYOUT or MODULE_CODE_TRIPLE after ResolveDataLayout runs.
> > > I guess we could test that?
> > That would require crafting manually some invalid bitcode wouldn't it?
> Yes, or better, something we now call invalid. @efriedma please correct me but there could be an input we accepted before but not now? 
Yes.  But no tools exist to generate such a file.

I don't really want to sink a day into constructing a framework for BitcodeReader unittests, but if you think it's necessary, I guess I can.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D79900/new/

https://reviews.llvm.org/D79900