[PATCH] D98574: [Sparc] Define the same macros for -mcpu=v9 as GCC on Linux and the BSDs

[Brad Smith via Phabricator via cfe-commits]

brad added inline comments.


================
Comment at: clang/lib/Basic/Targets/Sparc.cpp:246-256
+  if (getTriple().getOS() == llvm::Triple::Linux) {
     Builder.defineMacro("__sparc_v9__");
-    Builder.defineMacro("__sparcv9__");
+  } else {
+    Builder.defineMacro("__sparcv9");
+    // Solaris doesn't need these variants, but the BSDs do.
+    if (getTriple().getOS() != llvm::Triple::Solaris) {
+      Builder.defineMacro("__sparc64__");
----------------
glaubitz wrote:
> glaubitz wrote:
> > ro wrote:
> > > glaubitz wrote:
> > > > jrtc27 wrote:
> > > > > This doesn't need changing, we can define more things than GCC to keep it simple.
> > > > Well, my original intent was to match GCC to make sure we're 100% compatible and I would like to keep it that way.
> > > I agree with Jessica here: you're creating a complicated maze for no real gain.  Besides, have you checked what `gcc` on the BSDs really does?  They often neglect to get their changes upstream and what's in the gcc repo doesn't necessarily represent what they actually use.
> > Yes, I have verified that GCC behaves the exact same way as this change and I don't see any reason not to mimic the exact same behavior in clang for maximum compatibility.
> FWIW, I meant GCC on the various BSDs. I do not think it's a wise idea to have clang deviate from what GCC does as only this way we can guarantee that everything that compiles with GCC will compile with clang.
> Besides, have you checked what `gcc` on the BSDs really does?  They often neglect to get their changes upstream and what's in the gcc repo doesn't necessarily represent what they actually use.

What is upstream is what we do. There are no local patches that change behavior in this particular area.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D98574/new/

https://reviews.llvm.org/D98574